Attack of the Robots - An Algo Journal

May 16th, 2019, 07:59 PM

kevinkdog

There is almost certainly a reason why you are missing orders, and it may or may not be the fault of Tradestation. I don't know.

BUT, what I do know is that any automated software solution will have its issues. So, changing from one software platform might just solve some issues, but bring up others.

The key with any platform you use is to perform root cause and corrective action analysis of any significant errors. I recommend you do just that. Then, depending on what you discover, you might conclude that a platform switch is appropriate.

I hope you solve it!

Order execution is really the least of our problems and we have had pretty good success on that front.

I suspect that the missed orders are a result of a memory leak or some bug in the Tradestation code. Running the same strategy against the same data set gives me two different results..... one set of results before restarting Tradestation and a different set after restarting. If I see it again, I will post an example of exactly what it looks like before and after a restart of the application. The strategy is clean.

Our Tradestation issues run far and deep beyond this particular issue and we have experienced a ton of problems over the years. I have installed and uninstalled TS more times than I care to think. Registry edits, lost data, corrupt files, installation failures.....you name it, we've done it with TS. We just don't trust the platform any longer, especially as we continue to grow our trading operation. I work with high end ERP systems, both functionally and technically, and what I'm seeing is definitely a bug. I don't expect a perfect system, but I do expect some level of stability and reliability. I have had TS freeze on me at least twice this past month, while in a live position.

I appreciate the encouragement. I know I can get through this, but as I've said in other posts, TS days are numbered in our shop.

~vmodus

May 17th, 2019, 09:14 AM

@kevinkdog , I'm sorry if I may have sounded grouchy in my response last night. I gave what you said last night some more thought. It occurred to me that I can/should use the print feature in Easylanguage to see what my code is doing.

I have the print feature written in all of my strategies and have a parameter to turn it off or on. Here is the code:

Code

// Easylanguage code snippet
inputs:  // just add the inputs below to your existing inputs
   print_on(0),                  // 1 = turn print log on
   print_start_date(1190101),    // start date for the print log
   print_end_date(1190101)       // end date for the print log
   ;

vars:    // these variables are just for reference
   iTrend(0),
   Trigger(0)
   ;
   
// Condition to print 
Condition90 = print_on = 1 and Date >= print_start_date and Date <= print_end_date ;

// if condition is true, then print to the log
if Condition90 
   then
      print(Date, "-", Time, " - ","iTrend: " , iTrend, "Trigger: " , Trigger) ;

I actually add more 'print' statements as needed, to see other things as well.

Here is what the parameters look like on one of my strategies:
[img]https://i.postimg.cc/bJYWztz3/SNAG-0225.png[/img]

So today I'm running it in sim again this morning, with print activated. Thanks again.... you got my thinking turned in the right direction. I'll let you know if I need assistance.

~vmodus

May 17th, 2019, 11:02 AM

My weird trading thought of the day:

I am working primarily with the VX tool, and I'm pretty much in an upside down world. Like most people here on FIO, I don't care which way the market moves, as long as it moves. The weird part is that, as I'm watching my instrument fall (short), it is actually good news for the markets. So I cheer for my instrument to crash, this means the market (specifically the ES) is moving up.

Just a weird thing to experience.

~vmodus

May 17th, 2019, 04:42 PM

So today's sim run worked a lot better than yesterday. Using my lessons learned, I started TS fresh, only one strategy and nothing else (except trade manager once trading commenced.

Of course the results were not the best:
[img]https://i.postimg.cc/RCG3YSkg/SNAG-0227.png[/img]

From a market standpoint, there were three ranges that got to us. We were whipsawed a little at pre-market and then just at open, got in the correct side for most of the morning, then some whipsaw action, then another long trade, then a little more back and forth.

Today's Results:

-$764
5 trades, 3 losses and 2 wins
Compare to the strategy performance report: -$867

So we did a little better than predicted by the strategy performance report. Obviously this was a losing day, but this week, for this strategy, would have been flat.

What I liked

Strategy identified all potential trades
Our actual P/L was close enough to the strategy report
Signals were good
Code did not fail like yesterday

What I didn't like

Strategy missed one trade, because of a timing issue; the limit price was not hit and thus we didn't get in.
There still work to be done on the trades that get missed
For this market and strategy, true automation might be difficult to accomplish
I pulled the plug on my strategy 4 minutes early.... I thought my EOD exit was not being hit.... then realized it evaluates at the end of the bar, thus I should have waited until 1610.

The VX is one challenging beast. I like the challenge, but the learning is painful. This strategy needs more work and I will have to think about whether I will trade this next week in sim or live. I am leaning towards sim, but will make a decision by Sunday night.

Anyhow, that is all I have for this week. I hope you all have a restful weekend and let's do this again next week.

~vmodus

May 17th, 2019, 04:57 PM

Fascinating stuff as always, do you think it is odd that a volatility index like the VIX goes down when prices go up in a volatile fashion? I need to look to see if there is a gauge for volatility that disregards (or is neutral with regards to) direction and only focuses on the "spikiness/variance" of the movement itself.

Have a great weekend!

May 17th, 2019, 05:33 PM

The VIX is skewed because investors in equities are typically long-only, and when fear rises, PUT prices rise faster than the drop in CALL prices, increasing the VIX, and when the market is going up and doing well, fear subsides so PUTs go down in price faster than CALL prices rise, so the VIX drops. This is why the VIX usually rises in down markets and drops in up marketsx.

In short, if people are bullish they generally buy underlying instruments like futures, ETFs, and stocks, instead of CALLs, and use PUTs to insure them.

Also, an underlying can't go lower than zero, but it can theoretically go up to infinity. This absolute-zero floor also makes puts more expensive than calls.

If you want to just measure volatility in terms of actual up/down movement, just use ATR.

May 17th, 2019, 06:24 PM

shodson

The VIX is skewed because investors in equities are typically long-only, and when fear rises, PUT prices rise faster than the drop in CALL prices, increasing the VIX, and when the market is going up and doing well, fear subsides so PUTs go down in price faster than CALL prices rise, so the VIX drops. This is why the VIX usually rises in down markets and drops in up marketsx.

In short, if people are bullish they generally buy underlying instruments like futures, ETFs, and stocks, instead of CALLs, and use PUTs to insure them.

Also, an underlying can't go lower than zero, but it can theoretically go up to infinity. This absolute-zero floor also makes puts more expensive than calls.

If you want to just measure volatility in terms of actual up/down movement, just use ATR.

Ahh thank you. I have never thought of volatility-skew in this fashion, I probably mis-understand it. I always think of it as merely "markets tend to crash faster than they rally" but the option-pricing is skewed as well. Thanks!

May 20th, 2019, 04:39 PM

snax

Fascinating stuff as always, do you think it is odd that a volatility index like the VIX goes down when prices go up in a volatile fashion? I need to look to see if there is a gauge for volatility that disregards (or is neutral with regards to) direction and only focuses on the "spikiness/variance" of the movement itself.

Have a great weekend!

I think it is better to think of the VIX as a fear gauge than a measure of volatility. Volatility can exist and the VIX (or VX, which is what I'm trading) can be low. For example, if this trade war were to be resolved, you would see the VIX drop like a stone, but see a ton of positive (bullish) volatility in equities.

Since the VX is tied to options rather than the underlying equity, it is related to options activity. I've oversimplified it, I'm sure, but I think that's the crux of it. Still learning about it as I go along here.

as @shodson mentioned, ATR is a good measure of volatility.

~vmodus

May 20th, 2019, 05:13 PM

Well, today was a good day. I decided yesterday that I would trade for at least one more day in sim. Today was the first day that sim actually matched my strategy performance report. Everything is spot on, so it is a very nice place to be.

Sim account, closed P/L = $188:
[img]https://i.postimg.cc/pXfwRsrP/SNAG-0229.png[/img]

Strategy Report (what would have been in a perfect world): P/L = $188
[img]https://i.postimg.cc/6qjGvr5q/SNAG-0228.png[/img]

Trade Manager Report, 5 round trip trades, no cancelled or missed trades:
[img]https://i.postimg.cc/4NyWFwc8/SNAG-0231.png[/img]

Today's Results (sim):

5 round trip trades (S, L, S, L, S)
+$188 net profit

I believe I have worked out most of the bugs. My end of day exit worked as planned, unlike last week when I was pulling the plug too early.

Final thoughts for today. Here is today's VX chart:
[img]https://i.postimg.cc/J0xjQyMD/SNAG-0232.png[/img]
There are generally 2-3 good moves in this market per day. Unfortunately the best move of the day happened at 4:40. Do I want to trade at that time? Probably not. My partner tried it and didn't sleep well for six months. I would really have to trust the execution of my automated strategy to just let this run overnight. I'm far from that point. Just some things for me to think about.

Tomorrow:
I might trade live, but I think I want one more day of sim.

That's it. Algo should be sweet and simple like this. I didn't do any trading work, other than restart my PC and login. And update this journal.

See you all tomorrow!

~vmodus

May 20th, 2019, 05:34 PM

shodson

The VIX is skewed because investors in equities are typically long-only, and when fear rises, PUT prices rise faster than the drop in CALL prices, increasing the VIX, and when the market is going up and doing well, fear subsides so PUTs go down in price faster than CALL prices rise, so the VIX drops. This is why the VIX usually rises in down markets and drops in up marketsx.

In short, if people are bullish they generally buy underlying instruments like futures, ETFs, and stocks, instead of CALLs, and use PUTs to insure them.

Also, an underlying can't go lower than zero, but it can theoretically go up to infinity. This absolute-zero floor also makes puts more expensive than calls.

If you want to just measure volatility in terms of actual up/down movement, just use ATR.

Thanks for the insight, @shodson. I did a lot of studying on options about 20 years ago, but never needed to know much until I started working with the VX.

~vmodus

Attack of the Robots - An Algo Journal

Discussion in Trading Journals

Attack of the Robots - An Algo Journal