Thursday, November 10, 2011

Making SIM Results More Real - An Experiment

On one track...
I have the plans to help me achieve the state of peak performance.  That's going to take me a lot of time to reprogram my brain, but based on a great comment from Flowtastical, it can be really as simple as: 
Study your best trades and execute them better.  That is peak performance.
On the other track...
I'm still trying to "make it my own" trading the ES with the DS system.  And since I'm back to trading on SIM (simulation account) to really nail down what my style should be, one thing that has been bothering me is the lack of confidence I have with fills on SIM with limit orders.

SIM fantasy fills - to good to be true
When you use limit orders to enter and exit a position, you usually get a fill even if it doesn't trade through your price -- I call those "SIM fantasy fills."  So today, I began a day of experimentation trading the DS system with more of an aggressive scalping style (12 to 30+ trades a day) so that I could generate as many trades to evaluate.

The goal - make SIM results more realistic
Since I was a bit taken aback by the differences between SIM results vs. live trading in my last foray, I wanted to able to find a way to trade my SIM account so that it would better match real life results.  Next time, I don't want to be so surprised about my live P&L differences vs. SIM.

Adjustments made when trading SIM
For my experiment, my profit target was set to +2.00 and stop loss was -2.00.  However, as I started to complain about how the SIM fantasy fills were too easy and not like reality, CLETrader (the resident expert in the DS room on scalping strategy) recommended some of the following revisions to my bracket order on SIM:
  • Move my profit target limit order to +2.25 (and including the bullet below = +2.50)
  • Place my limit entry at a 0.25 favorable (more difficult to fill) level
  • Stop loss would remain at the original price
  • A $5 commission added per round trip


The adjustments accomplished a few tasks
  • The original entry limit order would have been filled in real life
    • However, every order, regardless of outcome, will have overstated profit by a tick.  This needed to be adjusted in the final results.
  • The original profit limit order would have been filled in real life
    • However, the trades the hit the profit target order will have overstated profit by a tick.  This needed to be adjusted in the final results.
The downside - manual calculation work necessary
The only difficult part of this task is the additional overhead necessary to calculate and apply the adjustment for every individual trade manually or semi-manually via Excel to generate the revised results.  But to me, this is a task worth doing in order to give better confidence whether the trades would have filled (or not) in real life.

An error in my experiment
In hindsight, I applied the incorrect values to my bracket orders today, and didn't modify my profit target limit order by another tick.  In some cases, this did make a difference on whether I got the fill or not.

Experimentation results
Please note that I was NOT trading a specific strategy consistently throughout the day (although it was more scalping related), and in addition, note the error explained just above which also impacts the results.  So this "experiment" has a lot of holes, and the actual P&L is somewhat meaningless.

Cumulative P&L
However, focus on the gap between the raw SIM results (red dashed line) vs. the adjusted live results (blue line).  A quick glance helps to explains one of the reasons why some users (mostly active daytraders who primarily use limit orders) of SIM are shocked by their results when they start trading live.  As the day progressed, the chart illustrates how small adjustments began to compound the difference between Adjusted Live and SIM results significantly.

I had 36 trades (I'm really tired now!) with a not that great 51.6% accuracy, which goes to show how I wasn't on track with the DS system which usually gets 70-80% accuracy.  Average gain was $96 and average loss was $87 (both raw SIM output), so we're talking about some tight margins on this particular day. 

One more interesting highlights is how the raw SIM data ended the day showing a gross profit, whereas the adjusted data showed a significant net loss.  With a relatively large volume of trades, the commissions and adjustments quickly overcame any gross profits.

The takeaway
When trading (especially active traders) via SIM utilizing primarily limit order entries/exits with an even risk/reward ratio, it is very useful to make adjustments to your final SIM results to avoid a false sense of confidence and success.

The method discussed here of adjusting the entry orders, in addition to penalizing the results of the trades, is likely to error against the trader's P&L greater than reality.  However, if reality ends up being a little easier because of this approach to trading on SIM, I consider that a successful outcome.

6 comments:

Tomross58 said...

Hey Grove,,

Wow,,,I didnt realize how that can impact so much when you go live. When I started trading stocks I was on sim also at first and then just never realized after going live those fills usually arent in ones favor. ( Of course this isnt the only reason I wasnt successful trading stocks but thats another story,,)
I have never used a tick chart before and this would be a good discussion to have with Rena on entries and timing with the tick chart. When I was trading stocks I would wait for a level of support or resistance and as price arrived there I would wait for the top wick / bottom tail to form maybe 10 or 20 cents and use a market order with my stop above or below that wick. I actually got pretty good at it and am sure that I didnt get the best entries but usually within my risk parameter. Sometimes I would get stopped out and price would head fake and I would try a second time but after that I would let it go if it did it a third time. I wonder if this is possible trading the e-minis and useing a tick chart.
I really think you are working on something very important here as I do beileve that stops will preserve ones account but order entry is such a key element. How many times do we get the direction correct but are just a bit early? or how many times do we miss our entry by just a tick only to watch it take off in our direction. This brings me back to the thought of not being perfect but being good. There is a way that we can make good entries that overall will give us an edge,,,,,,at least I think so.

Tom

MBAGearhead said...

Very good SIM approach. I was going to suggest the same thing. Personally, I hate SIM because you are missing 50+% of reality. The mental side is what messes up most traders, myself included. You just can't experience that with SIM. Good luck!

Grove Under said...

Hi Tom:
Great point about using tick charts for entries, since this is something I have never done before. I'm really starting to like these type of charts, although it has pros and cons just like any other chart type.

I've spent the past few days experimenting a lot with the tick charts, especially for entries. And based on your comment on timing and entering based on a market order, I believe it's possible, but for the ES, I believe it's pretty tough to consistently get a good fill.

This is because during turning points or support/resistance levels, the ES can move fast, and literally a 1/2 second (or less) delay of the mouse click could result in a fill that's a few ticks off from where you expected. And if your profit target is 8 ticks, that slippage is a huge percentage of your potential profit.

What I have found is that by using the tick chart in combination with higher timeframes, you really can start to pinpoint specific levels to place your limit orders. And those levels from the DS system seem to have better odds for a turning point vs. other more random levels.

Thanks for the comment, still lots for me to learn for sure!

Grove Under said...

MBAGearHead:
You're so right about how trading on SIM leaves a lot to be desired!

I'd have to imagine that the mental component could be more like 80+% of reality. So unless there are some tricks you can play on your own mind to make it seem more real, the gap between SIM and reality can be quite wide.

After using the new "tougher" SIM criteria today, it really did start to feel more "real" (i.e. missed many fills on entry and profit targets), but there's still something about not having real money on the line that changes everything, no matter how hard I try.

So I might have to return to live trading sooner rather than later (but trade only one contract), just to be able to have the most realistic conditions upon which to learn.

Thanks for the comment!

Greg said...

I know this post is old now :), but I had an idea. Probably part (most?) of the sim vs real difference is emotional, as opposed to fill-related. I could be wrong as you're talking a lot in this particular post about the mechanical parts, but let's say for the sake of argument that it's emotional.

The problem is, there's of course nothing actually on the line when trading sim. We can say all we want that we treat it the same and such, but there's nothing like having real money on the line to change our behavior.

So my random idea is to turn the sim trading into a game with real stakes. The goal is to get back to trading real money. I'm making up these numbers, but something like:

1. Start at some baseline sim account balance.

2. At the end of the day, take 20% of your sim P&L (positive or negative), and add/subtract that from your total "available risk amount" in your real account.

3. Trade your real account whenever you want to, but only risk up to the available risk amount.

So say you trade sim for a week, and end up +$3000 in the sim account. Based on that, add $600 to your available real-account risk. If you trade for real then, go for it, but if you get below -$600, you're out.

Similarly, if you lose $3000 in a week, subtract $600 from available real risk, which means you probably have to make that back again before you can trade.

I'm making all this up as I go...but my general thought is to put real-ish stakes on the sim trading. You really want to trade your real account, so put your ability to do that on the line when trading sim.

Anyway, just a thought!

Grove Under said...

Hi Greg,
Great point that much of the differences between live and SIM is emotional. I totally agree with you on that.

You have a very interesting example of incorporating SIM with your real live account. This way, you're "earning" your way to trading live, so it means a lot more to trade well via SIM.

I'm still trying to figure out the criteria for when I go back to the live account, so this is timely and helpful. Thanks for sharing this idea, much appreciated!