I would probably scrap Strategy #1, because there is a 99% chance they wouldn't let you trade it Live. It is overnight trading and my understanding is that they probably only let very senior traders to do that, if at all. It also doesn't give too much to your bottom line, so unless you just want to test it (but you already did that) there is no point in running it.
Also you could go over the other strategies and see if they somehow violate their rules (or rather preferences), like averaging down for example. Again, averaging down is a viable strategy if used in a restricted way, but they wouldn't let you trade that way Live, so there is really no point in using such an approach in a Combine...
Otherwise good luck....
The following user says Thank You to Pedro40 for this post:
Good luck on your next combine!
Will you be trading 6-10 contracts per strategy or 6-10 contracts total anytime of the day?
In other words it was not clear to me if multiple strategies could have working orders at the same time...
The following user says Thank You to gulabv for this post:
Good points. No averaging down for me, so that is not an issue. After I developed strat #1, I found out that overnight trading is discouraged, except for senior traders. The primary purpose of strat #1 is to improve the overall winning percentage, so if in later stages I can't trade it, that will be OK. My winning percentage will just go down, and maybe profit will decrease a bit.
Thanks for asking. For strats 1-3, they are all basically traded at different times, with no overlap. So each trade will be 6-10 contracts. The tricky part will be when I add in strat #4, which trades same time as strat #2. I probably will treat them as independent trades (and size them accordingly), but I will have to keep a close eye on the Daily Loss Limit. Some of the details will therefore be decided in real time.
The following user says Thank You to kevinkdog for this post:
Since I started Combine #2 tonight, after this post I will only be posting those results.
So, here is the final chart for Combine #1 (thru trading day 20), and the month or so since then (up to June 15, 2013).
I will still be tracking this for my own benefit, since I may trade these with real money soon.
For what it is worth, if I had to decide today whether or not to trade these with real money, I'd probably say "not yet," even with the excellent recent performance. Why? Well, for the first 25 days, the combined strategies underperformed my expectations by quite a bit, and for the past 10 trade days, those same strategies have radically out performed expectations. I don't like trading based on extreme behavior like this - I'd like to see it more closely mimic the navy blue "average" line.
So, I'll give it more time, and see what happens...
The following 5 users say Thank You to kevinkdog for this post:
It is always good to start a Combine with a positive day.
Needless to say, I started Combine #2 with a negative day. Great way to start!
I also had a couple of trade entry/exit issues, so I did a little worse than I should have.
With only about $2,500 left before the max drawdown is hit, I will be trading only 3-4 contracts for the next few trades, until (and if) I get some cushion.
The following 3 users say Thank You to kevinkdog for this post:
Relative to the Max Drawdown metric , I would agree with you.
I'll explain how I came up with my initial size.
My initial size of 6 was a calculated risk I took, based on the Profit requirement, Daily Loss Limit and Max Drawdown. Instead of all metrics for the $150K Combine being 5x of the $30K Combine, they are 3x (max DD), 6x (Daily Loss Limit) and 8x (Profit). That means to use my existing strategies (which were designed for the $30K Combine), I had some position sizing decisions to make - and I knew they could backfire on me.
My initial size was 6 contracts. This is based on the higher Daily Loss Limit for the $150K Combine, which is 6X of the previous $30K Combine ($3,000 vs. $500).
The downfall to this logic is that the Max Drawdown for the bigger Combine is only 3X larger, not 6X.
I took all this into account, and determined if the first trade was a loser, I'd have to scale back. Of course that is what happened. So, the next trade will be 3-4 contracts, and if that loses, maybe even smaller size (depending on the next trade's required stop loss).
The plan will be to go back to 6 contracts if/when the overall account gets back to breakeven.
The following 2 users say Thank You to kevinkdog for this post:
It is pretty much as many contracts as I can trade, and still respect the Daily Loss Limit (DLL) and Max Drawdown (MD). I am trying to maximize profit, given the DLL and MD constraints. So it is not a fixed fractional based on account equity, but it is based on DLL and MD.
So, for the first trade, the stop was around $450 per contract (which I assume could hit $500 after slippage). That is also my max stop loss ever on any trade with any of the strategies.
Daily Loss Limit = $3000 divided by stop loss $500 = 3000/500 = 6 contracts (note that this will be the maximum size, regardless of future performance. As long as my stop on a trade is $500 per contract, I can't go above 6 contracts and still meet the DLL rule).
Tomorrow, the max Drawdown will be the limiting factor, since I have $2500 left. If I allow a $1500 loss fo the next trading day, that will still leave me $1000 cushion.
So, 1500 / 500 = 3 contracts for the next trading day.
If I lose the next trading day, I will go down to 1 or 2 contracts. If I win, depending on the winning trade amount, I will attempt to increase my size the next day.
I typed this fast, so if it is unclear, please feel free to question my logic. It is tailored to the Combine requirements, and that makes it a bit strange.
The following user says Thank You to kevinkdog for this post:
Right now, things are going better than expected. Long term, performance should revert to the average line. So, I consider the current performance a "hot streak" that will not last.
I was away for a week at Disneyworld, so I was not able to update the performance of my 3 strategies until now.
Right now, things are going better than expected. Long term, performance should revert to the average line. So, I consider the current performance …
And that, my friends, is precisely what happened in the past 3 trading days (including today). The 3 strategies are reverting back to the mean.
Normally, this would not be a problem. It happens, and I fully expected it.
BUT, it just so happens that I started the Combine #2 right at the peak, and it has been nothing but losers ever since!
If I was a discretionary trader, I'd be convinced there was something wrong with me - that the Combine was messing with my brain and impacting my trading:
A. Trade Combine #1, barely break even. Everyone can see the reports. B. After Combine #1, my performance took off. I'm trading great! No reports, so only I know the results. C. Combine #2 starts, trading is awful. Everyone can see the results.
That sure makes it seem like the results are due to me. Or that I am cheating. I know that is what I would think if I saw someone else post it ("Sure buddy, it is just coincidence that you trade great when no one is watching, but when your results are tracked by a 3rd party, your equity curve just goes down. Sure, just coincidence...").
But, since I am trading mechanical signals here, I can assure you the results are 100% legit. It is just some incredibly bad timing...
The following 6 users say Thank You to kevinkdog for this post:
thanks for the journal. Very encouraging to see mechanical system in the making, hopefully gets me going too.
I've been sort of developing a mechanical trading system and have been studing the anatomy of losing streaks through lower time frames. I guess I am trying to ask if you have done such a study in a way that you could reduce or eliminate your losing streak and/or turn the losing streak more into a flat line in your equity curve rather than a steep drawdown?
Regards,
Tihfa
The following user says Thank You to tihfa for this post:
Here is my opinion on your question (please realize others may have valid opposing viewpoints)...
When you create a system, look at the equity curve, and then spend a lot of time trying to improve the weak spots (drawdown periods, flat periods, etc), you are starting to descend a slippery slope. By that I mean it is fairly easy to make backtests better, but it comes at a cost - more rules, more parameters, more filters, etc.
What makes you think those additional rules will work in real time? They may, or they may not. But your mind will think there will be an improvement, since your backtest is much improved. This can trick you into thinking you have something "better" when in all likelihood you really have something that probably is worse.
It is also a slippery slope because once you do it the first time, now another part of the equity curve will look poor. Do you try to fix that, too? How many times do you repeat this process? Done enough times, you'll have a great looking backtest, but when you go to real time, the system will probably fall apart.
My general experience is that you can make adjustments like this 1 or 2 times to a strategy, as long as you leave plenty of data out of sample. Beyond that, it is probably better to just drop the strategy.
Just be very careful.
The following 2 users say Thank You to kevinkdog for this post:
This doesn't only apply to an automated strategy; a fully discretionary trader has exactly the same issues, just expressed a little differently.
Say you are looking at patterns, or trendlines, or levels, or some signal, anything that you look at over time to decide whether they are "reliable" -- you've got exactly the same situation. If you fine-tune to the past, you only know for sure that it "would have worked" during that period, not so much about any other time (like your next trade.)
So you hope you find something that is general enough that it applies to the market all the time; you have better odds of that, the simpler it is and the less tinkering you've done....
The following user says Thank You to bobwest for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
And this is exactly why traders that seek algorithmic trading as a way to solve their own shortcomings on psychology or lack of understanding in the market will still fail.
They think they can program their way out of their deficiencies, but your post #315 shows a good example of how discretion is still the #1 rule in the game. And that means you have to actually be good at it.
Trade first, program later. That's my recommendation for all. Once you are a consistently profitable trader, then you can start to program strategies.
BTW Kevin, none of this is aimed at you which is why I didn't quote your post. I am just speaking in generalities.
Yes, I've heard people say that algo trading is a way to take emotion out of the trading equation. Not true. Any psychological or emotional hangups one brings to the table with discretionary trading will NOT go away with algo trading. The still present hangups will just manifest themselves differently.
The following user says Thank You to kevinkdog for this post:
Hi Mike -
Can you explain this comment in more detail? Are you referring to maybe turning on and off an algo system via discretion, or that how discretionary systems could be a ton better than this algo system, or something else entirely?
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
Kevin, I meant the discretion in when to turn a system on or off, when it starts or ends, whether to stop it after a run up and wait for a pullback, whether to continue it after a drawdown, etc etc.
All discretionary decisions and ones that traders will deal with, usually in the wrong way, in my experience. I think this thread is great and that it shows how traders should be dealing with all those issues, more or less collectively called consistency, and how to measure it.
You bring up a great point: When is a good time to start trading a system (mechanical or discretionary):
A. At or near a new equity curve high
B. After a pullback in the equity curve
Both cases can be influenced by recent performance, too: a pull back after so-so upward performance is certainly different than a pullback after tremendous performance.
So, in my case for the Combine #2, I knew performance would eventually pull back - I just did not know when that would occur. Maybe it would be the next trade (which is what happened), but maybe the outstanding performance would have continued for the next month.
It is a personal decision, when to start. For me, I'd be more upset standing on sidelines, watching my system outperform. So, I started trading immediately. Others may be more risk adverse, and wait for the pullback. No right answer, no wrong answer.
But a key is not to second guess yourself. In my case, in hindsight I made the wrong decision, but it does not bother me. I take full responsibility for the decision, acknowledge it, and move on. I make wrong decisions all the time, so if I always second guessed myself, I'd be in serious trouble!
The following 2 users say Thank You to kevinkdog for this post:
The flip side to the argument about when to start trading a system is that "it should not matter." That is, be mentally, emotionally and financially prepared for the worst case, based on history. If you know the system once had a $5,000 per contract drawdown, assume you will start out live trading with the same thing.
Can you handle that worst case? Most people can't. I know for the Combine, I certainly can't, because of the max drawdown rule. With real money, though, I could handle it.
I bet many people give up way too early on decent or good trading approaches, purely based on a bad (but easily statistically possible) run right at the beginning...
The following 4 users say Thank You to kevinkdog for this post:
This morning my Strategy #4 gave a short signal. Problem was I could not take it in my Combine account (I took it in my personal account), because of the risk involved. For that strategy, I trade pairs of contracts, and if I lost on the trade, it would have brought me perilously close to the max drawdown point (end of Combine). So, I did not take the trade.
Of course, the trade never saw more than a tick or 2 of heat, and worked out pretty well (1/2 position still open, with $1,100 in total profits).
So, I feel bad I did not take it, but then again I don't, because my objective right now isn't passing the Combine, or making winning trades - it is surviving to fight another day.
I need to get some breathing room above the max drawdown ($145,500) so I can then shift into 1) trading more size and 2) switch into profit mode. To get there, I'd like to be able to endure 3 max losing trades in a row before I fail the Combine.
Right now I am at $146,618 in equity, $1,118 above max drawdown.
Before I go further with calculations, does anyone know ( @Pedro40 probably knows) if they turn you off at the max drawdown for open trades, or just on a closed trade basis? For example, let's say I have $145,501 in equity ($1 above "kill" point). Can I take a trade that loses $500 during the trade (which would take me under the max drawdown equity limit), but then closes with a profit? That answer might change how I work in "survival mode."
The following user says Thank You to kevinkdog for this post:
My strong suggestion would be to ask them directly; either an email to support, or, better, ask for a clarification from someone on the decision-making side, like John Hoagland. I think they would be willing to clarify it, and I think getting it officially would be important for you. (This is no knock against Pedro or anyone else; it's just that having it from the people who are actually going to enforce it is important.)
Good luck... you're managing this turn of events well.
The following user says Thank You to bobwest for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
Hmm.
I would feel bad.
I am just an outsider looking in, so take that with a grain of salt naturally. But, it seems to me the strategy was overleveraged from the start which is causing these issues.
Being forced to skip a trade because of risk management, account sizing, etc, is a flaw in the original strategy, in my opinion. And trying to cherry pick which trades are worth taking, trying to manipulate the size of the trade on the fly, etc -- well, none of it was part of the original backtest, based on my understanding. So you are greatly altering the results.
If I can be so bold, and hopefully not piss you off because I like you, this thread, and want you to come do more futures.io (formerly BMT) webinars --- I think the issue here is a bit of ego. You have a sense of needing to prove yourself publicly and don't want to fail a combine. This is affecting your decisions.
Normally, I am all for public accountability. Most people in that situation are struggling traders that need the accountability to stop making really stupid decisions (to be frank). But in this case, I think you are a good trader already, and are feeling some pressure to publicly demonstrate that, which is leading to some bad choices.
When I asked several weeks ago their response was that that only the daily loss limit will shut down your trading. The max drawdown is on an "end of day" basis so you can go below it intraday but once the day ends you must be above your max DD. Obviously you should check with them as they do make modifications to their risk parameters on an ongoing basis.
Seek freedom and become captive of your desires. Seek discipline and find your liberty. - Frank Herbert
The following 3 users say Thank You to Surly for this post:
Yes, you can go below the max. DD, as long as you bring it back by the close. I have done it more than once. If I recall, you can actually close the trade too, and start with a new one, as long as the DLL hasn't been hit.
Doesn't really make sense, but that is how it works....
The following 2 users say Thank You to Pedro40 for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
It has to simply be a loophole due to technology implementation on the reporting side. I would certainly imagine that a live funded trader, who has a live risk manager, would not be allowed to repeat the same thing.
Thanks for the comments. I do appreciate them, and I like seeing the perspective of readers on this Combine.
The Max Drawdown rule was, in hindsight, something that I did not spend enough time preparing for. Of course, now I am dealing with it, so we'll see how that goes...
I forgot to post Day 3 results, so here are Day 3 and Day 4 results. I have included the trade I did not take because of the risk of ruin in the "perfect" calculation .
Next trade will be 1 contract. It will increase to 2 for the trade after, if the win is big enough.
The following 2 users say Thank You to kevinkdog for this post:
I'd like to start a discussion on Risk of Ruin. The general concept of a Risk of Ruin analysis is to determine if you will go bust over a statistical number of trades. For example you may have a probability edge of 65% but after factoring commissions, …
Since you have a lot of experience in this area I would welcome your thoughts on that thread, if you think anything can be added to the discussion and spreadsheets.
2 trades today, 2 small winners. With both trades, I traded only 1 contract, per the plan.
Due to increased equity, next trade will be for 2 contracts. So, with my luck, it will probably be a loser!
Big Mike had mentioned a couple of times that he thought I was overleveraged, trading 6 contracts at the beginning. Sizing that big at the start definitely hurt me (it would have helped a lot had I had some winning trades, of course). To "normalize" the results, I include a chart at the bottom of trading 2 contracts for each trade. That should give everyone an indication of how position sizing is helping or hurting my overall performance.
So far, as Big Mike pointed out, my position sizing is really hurting my performance.
The following 4 users say Thank You to kevinkdog for this post:
Slowly clawing my way back. Next trade will still be for 2 contracts, and if it is a decent size winner, I will increase to 3 contracts.
As I had mentioned before the Combine 2 started, I am using position sizing that is designed to maximize profit, while still respecting the Daily Loss Limit. Plus, it now has to account for the Max Drawdown.
So, for each trade so far, the position size may have varied from 1 to 6 contracts, but in each trade I followed a set schedule for size.
Why do I mention this? Mainly, I want to be clear that position size decisions are not made in the heat of the moment (which almost never ends well). I recommend people make those decisions beforehand, when they have a clear mind and focus. Then, just execute to the plan.
The following 5 users say Thank You to kevinkdog for this post:
You have asked for comments so I look at this as if I were considering having you trade my account.
The first thing I notice is that your losing trades are bigger than your winning trades. To me that is lack of risk control.
The second thing I notice is whether you have the discipline to follow your plan.
On June 17 you wrote.
It looks like you traded 6 contracts the next 2 days. I’ll admit I just browsed quickly over the thread but did you comment on the change of plan.
Is there a planned risk reward ratio for your systems?
On a per contract basis so far you are risking more per trade than you gain. Your winning percentage has to be greater than 60% to break even. Right now, after 10 trades, your winning percentage per trade is 70%. Hopefully you can carry that forward.
That being said, you are now down to 14 days remaining. You have to gamble. To have any chance of making the required amount by the end of the combine you have to get your size up. At the rate you are going, a few hundred a day, you barely have a chance of breaking even.
I’m not familiar with the combine rules but does it make any difference if you hit your max drawdown or if you finish a few dollars short of your target? If it doesn’t then what do you have to lose? I think the goals for the combine are unrealistic. It’s kind of a go big or go home scenario. Best of luck.
The following 2 users say Thank You to deaddog for this post:
Thanks for the comments - it helps keep me accountable. Here are some replies...
I'm not sure why you say that. Would winning trades > losing trades imply I did have risk control? I don't think that is necessarily true. Can you explain?
If you read through the thread, you'll see I have followed the plan pretty much exactly. I have made mistakes, and the plan might have flaws, but I have followed it. I think the one thing I do have is discipline. Maybe I am wrong, though - where do you see holes in my discipline?
The trade report is a little deceiving, since it says "6" contracts and 1 trade. But TST counts in and out. So, one trade with 3 contracts would be 3+3 = 6. You can verify this by commissions, which are $5 per round turn. 5 * 3 = $15. So, I did indeed follow the plan.
No, since I am trading 4 strategies, all with different characteristics. Some are little profit, big loss (to get high win %). Others are big win, little loss (to get max profit). They all have positive expectancy, though.
The Combine is 20 trading days MINIMUM. So no need to gamble. I agree it will be hard reaching the profit target trading small size. Even as it gets towards the end, I won't gamble - what is the point? That won't prove my strategies are good - it will just prove I can violate my plan.
Yes there is a difference. If you violate the max drawdown, you FAIL. If you are above the max drawdown, but negative, you FAIL. If you are positive but under the profit goal, you can ROLLOVER (try again).
I agree the goals are VERY tough. That is why I took the challenge when TST approached me about a webinar. Even though I am doing poorly so far, I really enjoy the challenge of the Combine.
Thanks again for your comments. I hope I answered sufficiently. If not, just let me know,
The following 2 users say Thank You to kevinkdog for this post:
Relating to my own trading. I find that if I keep my winners greater than my losers I end up profitable. I’m neither a futures nor a day trader. I look for 1 to 3 risk to reward.
Like I said it was a quick glance and it was the first thing that I noticed. When ever I see a losing trader that is usually what stands out, big losses and small wins.
I also noticed that one of the additional performance requirements are for an overall average win being greater than the overall average loss. I would think this would lead to a consistence position size. I should also take the strategy that has a few big losses and many small wins out of play.
Now that I know how to read the report I don’t see any. I looked at the report and assumed that the 6 meant 6 contracts. You had said you were going to 3, which you did, and I was under the impression that you had traded 6.
How many days can you trade?
You might want to consider a gamble as you get near the end if you are not profitable. If you fail by not being profitable then what do you have to lose?
The plan for completing the combine profitably should probably trump following your trading plan. Depends on your objectives I guess.
Thanks Kevin; you cleared up a lot of questions.
Is there a reason that you use 4 strategies? Are there not enough set ups with just one strategy in the time allowed?
The following user says Thank You to deaddog for this post:
Understood. It really depends on the strategy, in my opinion. One option strategy I use has small wins, but big losses. The key is keeping everything is balance, to be overall profitable.
Many times, and especially for discretionary traders, having losers a lot bigger than winners does signal either 1) a need to be "right" or 2) unwillingness to admit you are wrong in a trade.
You could trade everyday. For me, it is a 2 month Combine. Gambling just to get rollover chance? Hopefully I won't have to consider that, but it is an option. Right now, I'm not inclined to do that.
I explain this early in the thread. Given the time constraints I was under, I could not find 1 strategy that met all requirements of Combine. So, I combined multiple strats to (hopefully) get there.
.... I have observed is that this ALWAYS backfires.
I am still of the opinion that good trading and adhering to the process of trading well keeps you afloat instead of trying to beat parameters. This is assuming that your process has already been tested and found to be adhering to the parameters of course!
And I can vouch for @kevinkdog's iron discipline - he is my role model as far as discipline is concerned!
The following user says Thank You to iqgod for this post:
Again it depends on your objective. Are you trying to pass the combine or at least not fail, or are you practicing discipline by sticking to your system?
The gambling is done at the beginning of the combine when you take a position large enough to put yourself deep in the hole on your first few trades. If it works out you are in relatively good shape for the remainder of the combine. However if it goes against you you have to dig yourself out of a deep hole. A calculated risk or a gamble?
If you have the confidence that you can dig yourself out of a hole if the trades go against you then why dig the hole so deep. Start with minimum size and work your way up. If you have to have the bigger size to pass the combine you can use it at the end. A calculated risk or a gamble?
The following 2 users say Thank You to deaddog for this post:
Calculated Risk: a trading action you take fully realizing the consequences, and being able to emotionally and financially live with the outcome.
Gambling: a trading action where you HOPE for a good outcome, and PRAY that the bad outcome doesn't happen. Also, any action outside of your designed trading plan. Or, worse yet, not having a trading plan!
So, in this Combine I have taken some calculated risks that did not work out. I've avoided gambling so far.
The following 4 users say Thank You to kevinkdog for this post:
Nice winning day today. I am now close to breakeven, so Monday we will start with 4 contracts.
So far, my position sizing technique has made my performance worse. That makes sense, since most of my losing trades had 3 or 6 contracts, and most winning trades were only 1 or 2 contracts.
The following 2 users say Thank You to kevinkdog for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
Yes, if you were to reprogram your strategy to duplicate your behavior here it would undoubtedly perform worse.
I've found that any momentum or trend strategy always performs worse with stops or dynamic position sizing (such as anti-martingale, decreasing risk on trades that go against you, increase on trades in your favor).
So question is, what will you do about it going forward - now that you are back to breakeven?
All this sizing variation is something I never do in real life. Then again, I never have an initial drawdown requirement like there is in the Combine. In real life, I usually size to maximize rate of return, given a maximum allowable drawdown (which is from the equity peak, not from starting capital like TST Combine).
Maybe I will attempt some different position sizing techniques in my simulator, and see what turns out to be best. If I do that, I will share my study with everyone.
As of right now, though, I intend to stick to the original plan, which will be to trade as big of a size as possible (once I get to or above breakeven), while still respecting the Daily Loss Limit requirement. That means a max of 6-10 contracts, usually 6.
If I go back under breakeven, I will cut size so as to not fail the max drawdown rule.
I'm not positive this is the "optimum" way to trade this Combine - i.e., maximize chance of passing the Combine while minimizing the chance of failing. I don't think I am far from optimum though, since I know trading 3-4 contracts for the whole Combine would probably never get me to a pass situation.
The following 3 users say Thank You to kevinkdog for this post:
So your risk as a percentage of your account is 0.28% per contract and if you trade 4 contracts it is just over 1%?
Your max drawdown is 3%. So 3 losers in a row with 4 contracts and you're out of the game?
You haven't mentioned strategy, but is there a reason why $425 per contract is your risk?
I ask this because I don't have a defined dollar risk when I trade. (Stocks not futures) I use a percentage of my account, determine the risk per share depending on where I enter and place my stop and calculate the position size based on the percentage of my account.
The following user says Thank You to deaddog for this post:
For the Combine, I did not look at it this way, since the $150K starting equity is really just a number, devoid of any meaning, since the metrics you have to hit are all dollar based, not percentage of equity based. The starting equity could be $1.5 million, and I'd still trade the same, since profit, drawdown and daily loss limit are dollar limits.
Sort of. First the max drawdown is calculated based on starting equity, not peak equity. So the requirement is really never to have closed equity drop below $145,500. I could theoretically run equity up to $300K, and then have a drawdown of $100K, and not violate the rule.
So, the 3 losers in a row would just be at the start, or any time I had $150K or less equity.
BUT, I'd never trade 4 contracts all the way down. I'd reduce to 1 contract to maximize my chances of staying in the game.
When I developed the strategies I use, I allowed up to $425 per trade per contract loss (leaving room for slippage and commissions, for a total per trade per contract loss less than $500). It turns out that $425 had the best performance.
Nothing wrong with that - it is a pretty standard method. I just had to do things a bit differently because of Combine constraints.
The following user says Thank You to kevinkdog for this post:
OK, based on comments I have received, and questions that I have received during my weekly Squawk Radio interviews on TopStepTrader, it is clear that many people think I used a poor position sizing approach for my current Combine. Before the Combine, I picked a certain approach that I thought was best, but I did not analyze it. Now I will analyze it, along with some others...
NOTE: This analysis is specifically for the system I am trading for the Combine. It might turn out that different systems will yield a different best result. So, don't take the conclusion of this analysis for your trading, unless you test it first.
PROBLEM:
Maximize the chances of passing the $150K Combine (need to hit $166K profit after 20 trading days)
SUBJECT TO:
1. Never let closed equity level drop below $145500 (the max drawdown rule)
2. Never lose more than $3000 in one day (Daily Loss Limit rule)
SOLUTION METHOD:
Try various position sizing techniques using Monte Carlo simulation (you can find this Excel tool earlier in the thread. I have modified it to give the various position sizing methods.) I input my daily trading results into the simulator, based on walkforward backtests. Results should generally be accurate, except it will not model days when there are multiple trades (and when I might use different sizing on each trade). This should not happen too often.
I will use 3 Position Sizing techniques:
1. Constant contract size, 1 to 10 contracts each trade throughout Combine
2. "Cut Size" model, where Combine starts with large size, and cuts down as equity gets close to max drawdown level (this is what I am using)
3. "Add Size" model, where Combine start small (1 or 3 contracts), then adds size as equity grows
Note: I am using my current position as the starting point for analysis. $148,900 equity, profit target $17,100, max drawdown $145,500 and 26 trading days left in Combine.
RESULTS:
Constant Contract Size Model
The table below shows the probability of certain events occurring. For example, If I trade 1 contract continuously, I have 0% chance of reaching profit target, 100% chances of staying above the max drawdown equity, and 100% chance of staying above the Daily loss limit. So, I'd never pass the Combine with 1 contract.
If you look at the results, I improve my chances of reaching profit target by increasing size, but only up to 6 contracts. After 6 contracts, I run the almost certain risk of violating the Daily Loss limit.
So, this says my best bet is to trade 6 constantly, but not more than that. I recommend you make sure you can see that before reading on. Ask questions if you don't.
Cut Size Model
Since in the Constant Contract Size model 6 contracts is best, that is where I will start. I will cut size, though, as equity dips below the starting equity. The idea will be to trade smaller as equity falls, so that I do not violate the max drawdown rule.
Here is the size schedule I used:
If equity > 150000 Then lotsize = 6
If equity < 150000 Then lotsize = 5
If equity < 149000 Then lotsize = 4
If equity < 147500 Then lotsize = 3
If equity < 146500 Then lotsize = 2
If equity < 146000 Then lotsize = 6 'last trade before failure, go for broke - if trade loses, size does not matter
Using this model, I can improve the odds of staying above the max drawdown, without too much of a drop in the odds of meeting the profit target.
Add Size Model
Since in the Constant Contract Size model 6 contracts is best, that is where I will end. I will add size as equity grows above the starting equity:
Add Size - 1 to 6 Model
If equity <= 150000 Then lotsize = 1
If equity > 150000 Then lotsize = 2
If equity > 152000 Then lotsize = 3
If equity > 154000 Then lotsize = 4
If equity > 156000 Then lotsize = 5
If equity > 158000 Then lotsize = 6
Add Size - 3 to 6 model
If equity <= 150000 Then lotsize = 3
If equity > 152000 Then lotsize = 4
If equity > 154000 Then lotsize = 5
If equity > 156000 Then lotsize = 6
Here are the results:
CONCLUSION
OK, so what is "optimum?" There is no clear cut answer, since it depends on what you feel is more important: A) passing the Combine profit target or B) staying above the Max Drawdown.
If you wanted to maximize your chances of staying above the max drawdown, you'd probably pick "Add Size- 1 to 6" Model, or maybe the "Add Size - 3 to 6 Model." But with those models, you have little chance of meeting the profit target.
If, like me, you wanted to maximize your chances of meeting the profit target, you'd probably pick the "Cut Size - 6 to 1" model or the "Constant Contract 6" model.
I had originally started the Combine with the "Cut Size - 6 to 1" model. Based on this analysis and my personal preferences, I still feel that is the best model to meet my goals.
EPILOUGE
Unfortunately, this analysis tells me that at this point I don't have good odds of passing the Combine. It also says that my choice of trading system is not the best for this size Combine (if you will recall, I designed this trading system for the $30K combine, where the performance goals are quite different). So, I tried to fit a square peg (my trading system) into a round hole (the Combine requirements) by using a hammer (my position sizing). It probably won't work.
So, what would I do, if I could start over? I'd modify the system to be able to increase size within a trade, or a trading day, to increase size after wins, without increasing initial risk. Basically, an "add to winners" approach. Having a smaller stop size (less than $425 per contract) would also be a big help, too.
Comments and questions are welcome, especially since I wrote this quickly and I may have missed some key points. If you disagree with my conclusion to use the "Cut Size 6 to 1" model, please speak up. Maybe I am looking at this the wrong way.
If you think you have a better sizing model, please present it in detail, and I will test it with my trading system. Keep in mind though that results will ONLY apply to my specific trading model. But maybe you can do better than me!
The following 7 users say Thank You to kevinkdog for this post:
Enjoyed your interview on TST, Kevin. Have you considered the issue that once you pass combine with your overnight systems you won't be able to trade them in LTP and Junior trader phases? They only allow to trade from 7 AM to 3:10 PM CST.
Actually, before I started Combine #2, I had a 30 minute conversation with John Hoagland at TST specifically about what, if anything, I'd have to change in my system from Combine to LTP to JT. His conclusion was that I could trade the same way. BUT, we did not talk specifically about overnight trades.
Given my performance so far in Combine #2 (and probabilities for passing given in my post earlier today), it is probably not something I have to worry about, anyhow.
For me, it looks like trading strategies 1-3 with my own money (I already trade strat #4 with my own money) will be the way to go...
The following 2 users say Thank You to kevinkdog for this post:
Have you considered adding size as equity shrinks?
Your aim with trading size off the get go is to maximize profit. What about minimizing your losses?
There is a calculated stop of $425 a contract. The starting size is 4 contracts. If you take a loss on the first trade then size is reduced. Lets say to 3 contracts, then 2, then 1 and there is a deep hole to dig out of. 3 losses in a row almost takes you out of the game. Even with a win on the 4th trade you are still in trouble. In the 4 trades you have traded 10 contracts.
Trading those same 10 contracts in reverse order; start with one and got to two if you have a loss, then 3 then 4. After taking 4 losses in a row you’ll be in the same position that you were if you start with 4 and work your way down to 1. The difference is that if you get a win in any of those 4 trades you are back to even or close to it.
I know that doubling down is a terrible idea for your own account but for a combine where you are out of the game after a very limiting max draw-down it might work. Minimum size can be increased as the account grows.
You haven’t addressed the risk reward ratio of your strategies. Are you expecting the wins to be much larger than the losses? Do you have stats on a per contract as opposed to a per day basis?
I will try your approach and probably post it tomorrow.
I can also post the Performance reports of Strats 1-3, and summary stats for Strat 4, so you can see the win and loss sizes. All the strategies are different, which makes it hard to lump into 1 Reward:Risk number...
The following user says Thank You to kevinkdog for this post:
I missed that this is how the draw down was calculated for the combine. I have been trying to calculate what sizing I should use to ensure I don't hit the daily loss limit and this changes the calculation once again.
Thanks for pointing this out
I also like your analysis on position sizing as I'm trying to work through this for my coming combine too.
Trade Wise, Trade Well
John
The following user says Thank You to MarketPilot for this post:
I haven't tried this approach, mainly because I do not know what exactly you'd like to do on the upside. Do you want to add a contract at every $1000 gain in equity?
Would you like me to try something like this? Please feel free to edit it to get your desired schedule.
It will depend on the R/R ratio to some extent. How much do you plan to make for your $425 of risk. It would also depend on the expectancy of the strategy and the frequency of the trades.
I would increase size when I could have 4 losing trades in a row and not put myself out of the game. Once you are over the initial capital you increase your size so that after 3 losing trades you are still in a profit position.
You go to 2 contracts at initial account size plus 6 times your risk per contract.
You go to 3 contracts at initial account size plus 9 times your risk per contract.
Once I increase size I would also change my position size calculation.
If I’m trading 2 contracts I would stay at 2 after the first loss the increase to 3 then 4.
If I’m trading 3 contract then I would stay a 3 after the first 2 losses then increase.
The reason I keep asking about R/R is that I’m assuming that the strategies employed will produce bigger wins than losses.
In theory with a 1 to 1 R/R you depend on the system having more wins than losses and requires either a large number of trades or more dollars be risked per trade. A R/R greater than 1 to 1 means you can take fewer trades or risk less per trade.
Kevin, I commented on this thread after seeing your performance on the fist few days. I’m new to futures.io (formerly BMT) and didn’t realize your credentials. I am appreciative that you are even taking the time to respond. Thanks for humoring me.
I'd like to start a discussion on Risk of Ruin. The general concept of a Risk of Ruin analysis is to determine if you will go bust over a statistical number of trades. For example you may have a probability edge of 65% but after factoring commissions, …
You are asking a lot of great questions, and it is helping me both understand my Combine position sizing better, and helping me explain things better. You've been a big help.
I hope readers realize that, at least for the trading strategies I am using in this Combine: 1) I have to do things I would never do with real money and 2) that in order to win the game (Combine), you have to respect the rules and make decisions based on the rules.
Some of the things I have to do with a Combine, that I don't do with real money:
1. Small per contract per trade loss ($425 before slippage and commissions)
3. Trade 1 strategy almost entirely because it has a high win percentage (in order to meet the Combine win % goal)
4. Trade another strategy in order to ensure I have 20 trading days in a 2 month Combine
5. Position sizing based on a very small initial equity drawdown (ruin point).
The list could go on, but you get the idea.
I guess the point is that you have to trade to the rules of the "game" whether the game is a Combine, trading your own money, trading in a contest, trading friends and family accounts, etc. All of those "games" will have different rules, and the key is to trade so that you don't fail any of them.
I tried to implement your idea the best I could. Here are results:
Your scheme does a really good job at avoiding ruin, but not so good a job at passing the combine.
So, I made one modification - I kept your downside scheme, but made the upside scheme such that once I had positive equity (>150K), I "floored it" and put on 6 contracts.
Here are those results. Not a whole lot different than what I am using (Cut Size 6 to 1). "Deaddog Modified" has better chance of not failing, but worse chance of passing.:
The following 4 users say Thank You to kevinkdog for this post:
Why are you limiting yourself in this manner? Increase your stop size per contract and decrease your position size. It's a percentage of capital at risk that should matter. Doesn't it come back to risk reward per trade?
Do you not have a daily loss limit trading your own account? How do you handle multiple losses on the same day?
. Trading a strategy that has small wins and big losses will probably get you in trouble considering the size you want to trade. Are you risking more than your calculated reward to show more wins than losses?
I would think you'd have to have a minimum of 2 strategies even in your own account. 1 for trends and 1 for chop.
Consider that this is set up to take other peoples money. Were it your money I was trading where would you draw the line. Isn't the first rule of trading to cut your losses short?
The thing with the combine if I understand correctly is that you don't fail as long s you are profitable. You don't pass but you are still in the game.
Thanks for the work. I guess I won't be applying for a combine anytime soon.
Firstly I'd have to master trading futures and secondly I'll wait until winter when I don't mind sitting inside all day. Might be an interesting way to test a strategy.
The short answer to your replies is that most of these things occurred because I created a brand new trading system specifically for the $30K Combine, which I am still using here, just with position sizing.
What makes you think those strategies will work on a 150K combine when they only had a less than 50% probability of working on a 30K combine?
With a 20 day 30K you are only required to make 7% of initial capital. With a 20 day 150K you are required to make 11% of initial capital. The requirements are 50% higher for the 150K.
The max drawdown is 5% of the initial capital for 30K and 3% on initial capital for 150K. Again the requirements for the 150K are much greater.
3 or 4 trades, 2 mistakes made which kind of balanced each other out. But overall another losing day...
I apologize for this thread turning into more of a joke ("like you'll ever pass this Combine! Ha Ha Ha!") than an inspirational journey to Combine victory...
BUT, there are still plenty of days left, and you never know what might happen!
Happy 4th of July everyone!
The following 6 users say Thank You to kevinkdog for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
Hey, it is no joke.
Here is the fact: in order to be a good trader, you have to be great at losing!
Even if you do fail this combine the thread is still full of extremely valuable information. And thanks to the research and number crunching you've done, you already knew your chances were less than a "sure thing" to pass the combine.
So I don't want you to get discouraged. First, you are helping many others by showing them "the other side" of algorithmic trading. And I don't mean the losing side at all --- what I mean is the research and analysis side that is outside and completely disparate from any kind of code to buy and sell. This is the side that so often is completely disregarded by most algo traders, and frankly the side that makes all the difference in my opinion.
So keep going, don't get discouraged, and good luck!
I'm glad you said that, because you've brought up some good points. I'm actually not discouraged - the raw system still shows a lot of promise (see below) - but I am disappointed in the Combine results.
I've been talking to a few Combine winners offline, and there is a lot of good info in winning Combines. Winning can be a more inspirational story than losing!
The following 2 users say Thank You to kevinkdog for this post:
Not surprised at all. The problem I see with the 2 contract position size is that the probability of winning the combine is probably very low.
Starting with a big position only works if you were fortunate to have a winner on the first trade, then you would have the cushion you need and be off to the races. Your gamble or calculated risk only pays off if the first trade is a winner. If not you have a deep hole to dig yourself out of and have to cut position size in order to stay in the game.
Possibly you should limit your position size to only increasing when you are won’t be negative if you lose. IE: you can risk $1000 only if you have a $1000 cushion.
Isn't there a fable about slow and steady winning the race?
The following 2 users say Thank You to deaddog for this post:
Kevin it would be interesting to add to these scenarios the calculation how long it would take to reach profit target, so we can get some visual interpretation of system profitability.
Trade to live. Not live to trade.
The following user says Thank You to xelaar for this post:
Number of trades per day varies from 1 (typical) to 4 or more (rare). If first trade of day is big loser, than there would likely be no trades rest of day.
I'm still debating in my mind if I should try to change contract size based on the TST Combine Draw Down from Initial Balance rule. I can easily see how after a winning trade you are enabled to then risk those banked profits to increase size on the next trade. Seems like here are the variables to better understand:
Profit amount needed from earlier trades to allow an increase in contracts
Relative change in contract number
Back Test Results on strings of winning or losing trades.
My trading style is all in scale out. So my added risk is that if I add too many contracts too soon, I could turn 1 small winning trade (at initial size) into the equivalent loss of 2 losing trades on a max stop out. I need to decide how to model it and run some calculations to better figure this out.
Trade Wise, Trade Well
John
The following 2 users say Thank You to MarketPilot for this post:
I have trouble with this.
Actually having trouble with the concept of dollars or risk per day as compared to dollars or risk per trade.
It sounds like you are taking a calculated risk (gamble?) that your first trade of the day will be a winner. By quitting if the first trade is a loser are you not missing the opportunity to trade whatever set-ups may appear the rest of the day.
Seems similar to the strategy to start the combine with large size.
The following user says Thank You to deaddog for this post:
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
OK.
Do you agree that there is no edge in trying to cherry pick which trade is going to be a winner or loser? So deciding to stop trading after 1 losing trade would seem to just be a random throw of the dice... the next 2 or 3 trades for the day could all be huge winners.
Of course they could be huge losers as well, but personally this is where I would look at the entire history of my backtest and find what "normal" is and then set my limits just outside of those areas a bit so that I would stop only if something "not normal" was happening.
I know you are probably tired of me talking about leverage. But this seems like another case of it working against you -- and you are making decisions based on something like a daily loss limit... but what do your backtest results tell you about that daily loss limit? That it is inside or outside "normal"?
Site Administrator Swing Trader Data Scientist & DevOps
Manta, Ecuador
Experience: Advanced
Platform: My own custom solution
Trading: Emini Futures
Posts: 49,733 since Jun 2009
Thanks: 32,283 given,
97,464
received
At this point I also feel compelled to write:
I am not saying I know how to do it better than you. After all, you are the one here posting in this thread, entered into the competition, and letting everyone look inside your system and critique your every decision.
Thanks for the comment. You have mentioned "cherry picking" a few times, and I'm not sure why the concept is relevant here. I have not used it in the Combine, or anytime else. I recall one day where I had to skip a trade, but that is because if that had been a loser, I would have been knocked out. To me, that is not cherry picking, that is trying to survive to fight another day.
To me, cherry picking would be passing on potential trades because I did not "like" the signal (I agree that is a fool's game), or cherry picking could be arbitrarily changing size of next trade (another fool's game).
What I have been doing is trading with maximum size whenever possible, to maximize the chances of hitting the profit goal. As I showed earlier, that approach maximizes my profit goal odds (at the expense of hitting the max drawdown limit), and has not worked (yet?) - my losers have had big size, and winners have had small size.
For casual readers, I should mention this: All this position sizing talk is because of the unique rules of the Combine.
For real world trading, I use fixed fractional sizing, with the "f" factor based on whatever maximizes the annual return / drawdown ratio. I also take every trade, all the time. Much simpler!
The following user says Thank You to kevinkdog for this post:
Thanks. I am pretty confident that, with the Combine rules and the particular strategies I am trading, I am making sound decisions.
I also realize that it is not working, but that is how this game goes. Bad things happen, even when you do the right things.
Repeat request: if anyone has a position sizing technique that they think would work better for my strategies for this Combine, please post it and I'll test it out. I'd love to know I have chosen a sub-optimal approach, because that means I'll learn something new. That is always good!
The following user says Thank You to kevinkdog for this post: