I am running a live account with TradeStation trading the Larry williams Oil Strategy. At the same time, I have developed some of my own strategies and would like to test and optimize them on TradeStation. Is there a suggested method to log into a demo account and test at the same time that I have the live account running?
The following user says Thank You to confusedxx for this post:
One example: you can backtest with a setting "fill limit orders only when price is exceeded" (as opposed to "fill if touched")
So if you backtest with this setting, you won't get limit fills unless your price is exceeded. Now say you run this strategy "live" without adjusting any settings. Going forward it will show limit orders filled as soon as the price is touched. But, if you refresh the chart, the trade will disappear since the backtest rules now apply.
I probably should not have brought up this issue, since the OPs question wasn't directly about this, but I get scared when it sounds like people are using SIM in Tradestation to evaluate a strategy. SIM is good for order testing, etc., but not for P/L evaluation.
The following 2 users say Thank You to kevinkdog for this post:
Thanks...makes sense. Personally, I see this being more of an issue with backtesting process, rules and rather than TS 'live' vs 'demo' mode and one being better/worse than another. That said...i mis-read your original comment to mean the TS demo account was 'bad' in some way due to data differences, etc.
The following 2 users say Thank You to ericbrown for this post:
I was actually referring to back testing to ensure code executed as expected and to see what I could possibly expect ad p/l given the markets I was planning to use. Following successful back testing, I wanted to run in sim mode. I do not want to risk real money with a newly crafted idea and strategy until seeing how it performs for a month or two in sim mode. I need to read the link provided above on this topic to learn.
How do you suggest I "live test" a strategy with live market data but on a simulated account ( basically paper trade)? I realize slippage and touching vs exceeding a price is a point of consideration, but I really want to test out my ideas on the best possible way in a live but simulated trading account. Finally, it seems that is also not possible with TradeStation. Meaning I cannot run my live strategies with real money and also use a simulated account in parallel? Any suggestions for how to better tackle this
The following user says Thank You to confusedxx for this post:
Yes you can. You will need an additional computer and an additional login (username) associated with your account. I did this almost 4 years ago when a friend of mine was trading my account live while I was developing/testing strategies on sim and I had 2 usernames - one for the live feed and the other for sim. You can find out how much it costs now to have an additional login but back then, I paid the platform+data fee for the sim feed and just data for the live one (platform fee was waived if you had 10+ round trips/month).
The following user says Thank You to Hulk for this post:
As I mentioned earlier, you can do it with 2 platform logins.
But the question is "will you get any benefit from doing this?"
Many times, Demo results have little correlation to real money results, meaning having a strategy that works good in demo mode may or may not work well with real money. I have run strategies in real time that fell apart, yet their backtest AND demo results were stellar.
The key is creating a strategy that produces the same results for backtest and real money trading. How do you do this? Really, experience is the key - knowing what you can and can't do.
For example, if your strategy uses 1 minute bars, and all you use to enter and exit are market orders, chances are your backtest and real money operation will be similar.
But, let's say your backtest includes tick bars and limit orders with small targets (ie scalping). Your backtest may look grand, your sim demo may look awesome, but you may lose very quickly with real money.
I always tell people a key to using Tradestation, Ninja, etc. is knowing the platform's limitations. I have just described one of Tradestation's. You can still be successful with it, you just have to learn its limitations.
Good Luck!
The following user says Thank You to kevinkdog for this post:
So after I create a strategy, I typically put it on the shelf for a few months, before trading it live. I'll check its performance once a month. I do this by checking its backtest result over the previous month. I compare that to my walkforward development test.
This approach works for me because I know what I program in the strategy behaves the same in real time as in backtest. This is the key.
Feel free to post or PM your code to me (take out any proprietary indicators) and I'll give you my thoughts on backtest vs. real money performance issues.
The following user says Thank You to kevinkdog for this post:
@kevinkdog has provided some good tips so far. I'm going to pimp his book and say that it'd be worth picking up and reading it. Great book for folks getting started in building strats.
Just curious, I have more than 10 round trip trades but Tradestation asked for platform fees for a second user account . I do not have platform fee for the primary account as long as I maintain $2500 in the account. Looks like they have different deals all the time.
They always have new deals going on. I found that if you submit your initial account application and wait, they'll eventually offer a 'deal' to get you to finalize the account.
I still question if you really need a 2nd platform. I've been using TS for around 10 years, I have 4 different accounts with TS, am running multiple strats live automated on multiple accts, and doing ongoing development and testing. All with one platform.
Kevin, how do you trade 4 different accounts with one platform ? I trade Futures, Forex and Equities from the same platform using the same user account. My intention was to test strategies from a Sim account during market hours while I trade using the live account.
Please let me know if there is a better way to accomplish this using Tradestation.
Simple, I can select the account I want to use if I trade manually, via the order bar. This is shown below. I can also select the account I want when I automate.
I know I can't run live and demo at the same time, but that is OK. I never use SIM mode - to me it serves no useful purpose.
So, my question for you is "why do you need to SIM a strategy?" What questions are you hoping it answers? There are likely better ways to accomplish what you want. SIM is notoriously bad in Tradestation.
The following user says Thank You to kevinkdog for this post:
SIM account is strictly for learning purpose. Was trying to test few automated strategies realtime. I understand I can backtest the strategies but was hoping to see this in action using a simulated account.
If you just have the strategy running, you can see how it would perform in real time (keeping in mind strategy fills prices won;t always be accurate). The only thing you would not see are actual orders being sent, and fills - that can only be seen in a Sim acct or in real acct. With the strat running, but not in any acct., you'd just see the strategy fills in your performance report. The SIM fills would likely be wrong (compared to real money) anyway.
If by "test few automated strategies realtime" you mean check that orders are sent properly, and fills and cancels come back right, you MIGHT see that in SIM account. But if by "test" you mean you want to evaluate P/L performance, SIM is definitely not good for that.
Hope this helps!
The following user says Thank You to kevinkdog for this post:
If you are using SIM to "prove" the profitability of your strategy, I think you are likely to reach bad conclusions. For example, SIM always assumes that limit orders are filled immediately upon touching. Real world that is almost never true. This means scalping type strats are almost impossible to SIM test.
What is the best way to test realtime? It really starts with your strategy. You have to create a strategy that can be replicated in the real time. For example, let's say you use only "buy/sell next bar at market" type orders in your strategy. If you include proper amounts for slippage (say 2x the normal bid/ask spread per round turn), your backtest results and live results will match up pretty well. This was the only way I did it for years.
As you start to get confidence that backtest and live are equal with simple orders, you can then begin to use more complicated orders. Then you can run limited tests with real money, and usually you will see pretty soon if there is a match (and if there isn't, a limited real money test won;t cost you a tone of money).
Good Luck!
The following user says Thank You to kevinkdog for this post:
Good to hear. Your results in SIM will then be a lot closer to reality. But I'd still be wary of the results.
I used to be more helpful at TS. I got beaten down by the paper trading trolls, so I don't post there much anymore. One troll actually shouted me down in a thread, only to apologize later via private PM. He wasn't man enough to apologize publicly. So, I post a lot more here, where things are civil.
The following 3 users say Thank You to kevinkdog for this post: