Nov 212010

After seeing how some other traders track their performance, I redid my spreadsheets.  I put all the data into one spreadsheet so I can see the performance over any time period, starting October 29 which is when I went back to real money trading after making the changes detailed in the A look back on my journey series (mainly dropping eminiwatch method in favor of FT71 method).  I also dropped crude, dax, & euro as I only traded them live for a few days and that was skewing the results.  For now I focus on Bund & ES and track them separately.

The way I was calculating $/contract was weighing all contracts equal and that wasn’t good.  If I scratched a trade with 4 contracts that would penalize a nice gain on another trade with only one.  It’d take the total gain and divide by 5.  That’s not really what I want, I want to track expectancy per trade on a per contract basis.  If anyone finds that confusing I can make a video on it.

Expectancy is win % * avg win/contract – loss % * avg loss / contract.  When I compute this for ES it’s 0.54 minus 0.10 for commissions so it’s 0.44 net.  That’s $22 or roughly 2 ticks.  I really need to get that higher.  I used to focus on profit factor but I think expectancy is much more important.  My ES profit factor is 5.04 which sounds great but a high PF is not going to pay the bills, expectancy will.

When I compute it for the Bund it’s 0.02 which is 20 euros.  I find it interesting that it’s about the same as ES.  This tells me I’m trading more for ticks than for the bigger moves.  And this is one reason why my average is so low.

The other reason why the expectancy is so low is because rather than stop out I’ve been scaling in on the next setup and then scratching the trade.  I mentioned that a bit on Friday.  I’m convinced this isn’t optimal and exposes me to additional risk.  I need to have a tighter stop, and when I stop out I can look for a second setup at a better price and then hopefully have a bigger winner than my first loser.  I need to test that this week.

I debate how I want to test this.  On one hand sim would be a good candidate but it’s hard to go to sim with a positive expectancy.  I’m averaging $250/day and I kind of like having an income, especially at Christmas time.  I could also continue trading 1-2 contracts as I am now and that’s what I plan to do.  Tightening the stop more will be a learning experience and as long as I can keep a positive expectancy then I’m good.  If my expectancy goes negative then I’d want to work things out on simulator.  So I’ll play it by ear.  This is a holiday week so I will be extra careful.

So here are two charts for ES.  The left axis is points and the right is each trade.  The bund charts look surprisingly similar:

The equity curve is interesting.  The first steep part up was when I got a nice mini-swing trades on ES.  I think I was motivated to do this to make up for some losses trading crude oil that day.  A bit of “revenge trading”?  I don’t know but I’m not trading crude with real money any more so I just focus on ES.

The flat spot (before the drop) was the Vetern’s day long weekend trip.  Then the steep drop was a loss I took on a long on Tuesday, the day the market dropped down.  Other than those incidents, the curve is pretty smooth.  Remember the graph is per trade not per day.  It would be interesting to track performance per day as well.

The expectancy graph shows the big trade as well (5 pts/contract).  I don’t include swing trades in these stats, just day trades.  I have a lot of 1 pt wins which makes sense as I usually go for +4 ticks & +6 to +8 ticks.  There are a lot of trades that make between 0 & 1.  Why?  One reason is if I get +4 on T1 and then T2 I get out +2 or 0.  That would make 6/3 = 2 ticks = 0.50 pts.  And the ones near zero were scratched trades.  So everything seems consistent with my trading.

Once I get the excel spreadsheets worked out I plan to share them here and hopefully get input from others.  I highly recommend keeping statistics on your trading.  I used to just use Ninjatrader but now I’m trading with X_Trader so I have to do it myself.  And I prefer it that way because I’m not limited by what Ninjatrader can do.  The only thing I miss is MAE and I’ll be tracking that (by hand) going forward.

So in conclusion, my focus this week:

  • Tight stops, minimize scaling in
  • Go for bigger setups.

I hope you found all this interesting.  The new journey to 100 contracts is underway.  So far it has taken a bit longer to get started than I thought.  I seem to be “stuck” at 2 contracts.  But I want my trading to be great before I start increasing size.

  14 Responses to “Measuring my expectancy”

  1. You are on your way. Regarding your calculation of an expectancy of 0.54 points for the ES instead of 0.25 as in one of your tweets, have you elimanted the scratch trades?
    I was surprised to read that FT71 has an expectancy of 2.5 points. Do you think that is per contract per day or per contract per trade. The latter one would be 10 ticks which would be very, very high for a short term trader.


  2. Thanks cunparis, I’ve always heard about expectancy but didn’t know how to calculate it (haven’t read Van Tharp yet). I went through my gap trades in October and computed an expectancy, in terms of ticks, to be much higher

    ES: +15.43 – 6 winners, 1 loser, avg winner = 14.667 ticks, 1 loser = – 20 ticks
    YM: +30.71 – 5 winners, 2 losers, avg winner = 23 ticks, avg loser = 50 ticks
    NQ: +43.25 – 5 winners, 3 losers, avg winner = 39.2 ticks, avg loser = 50 ticks
    TF: +10.00 – 2 winners, +7 and +13 ticks, no losers

    I think this is a very good metric to track.

    What do you use to track your trades? I have a custom spreadsheet I keep adding features to, but a friend of mine really likes TradingSpreadsheets and I’m thinking of moving to that.

  3. I do not eliminate scratches. And scratches have a big impact on expectancy. The reason my expectancy increased is because I’m doing it per trade. Before I was just summing up profit & dividing by contracts traded – this has the effect of weighting all contracts equal. What I’m really interested in is per trade expectancy. Some trades I use 1/2 position, some full, and some above full.

    2.5 pts is incredible because he starts scaling at 1 pt sometimes. Lately big trades have been more rare but I remember before he’d get 6 pts often. so 2.5 is definitely possible for him. Many times he’d hold a short for an hour or more and I’d be taking both longs & shorts (scalps) while he was holding short. This is how one gets a big expectancy, going for the best setups and staying in.

    When one trades 2 contracts it’s much harder to stay in. I scale out first 4-6 ticks and then I only have one scale out left. I usually take it at 2 pts. Once I am pleased with my results i’ll add a 3rd contract and go for 3 pts on that one. That would have a huge effect on the expectancy. But I want to improve results with 2 first..

  4. I am not sure if I understand the calculation. Is 0.54 your expectancy per trade but 0.25 your expectancy per contract?

  5. 0.54 is the expectancy per trade per contract. So whenever I place a trade, I can expect 0.54 per contract. If I do 1 contract then 0.54 and if I do two then 1.04 pts.

    0.25 (or whatever it was I can’t remember) is the result if I add up the total profit, say 30 pts and divide by the total contracts traded say 120.

    They’re not the same.

    Think of a situation where I go long 1 and get 1 pt. Then I go long 4 and I scratch for 1 tick per contract. Total profit is 2 pts divide that by 5 contracts = 0.20. This is how I was doing it at first.

    Now take the same trades and compute expectancy per trade per contract. First trade is 1 pt/contract. Second trade is 0.25 pts/contract. Now average those and you get 0.625. See how it’s higher?

    I feel the second way is more accurate. Because often I will scale in to scratch a trade.

    in either case, the goal is to raise it. Just averaging 2 ticks/contract per trade doesn’t leave much room for mistakes. I’d feel much better if it was 1 pt/contract. That’s what I’ve averaged in the past. 1 pt/contract per trade is pretty good actually. 2-3 trades would give 2-3 pts/contract. Increase size and it makes a really nice salary.

  6. PS: I’ll make a video going through the spreadsheet and then make the spreadsheet available for download. That would clear things up. I want to test it this week to work out any bugs and then release it.

  7. FYI, Ninjatrader backtesting results counts a scratch as a loss, because you paid commission and made no profit.

  8. FYI, I calculated my expectancy wrong, here is the updated, more realistic results

    ES: 9.71
    YM: 2.00
    NQ; 5.75
    TF: 10.00

  9. ‘”Think of a situation where I go long 1 and get 1 pt. Then I go long 4 and I scratch for 1 tick per contract. Total profit is 2 pts divide that by 5 contracts = 0.20.”. — You mean 0.4 points, I guess, but now I understand the calculation. The vid is a good idea anyway.


  10. Shodson – I count my expectancy before commissions to get the value (0.54 pts) and then I subtract the commission from this. $4.50 = 0.10 pts so that makes (0.44 pts).

    I’m not concerned about win % etc because a scratch that ends up +1 tick will be a winner and will ruin the avg win / avg loss ratio. A scratch at 1 tick loss lowers the win %. I just think these two aren’t as important. Expectancy & profit factor are more important IMHO.

    Are your results in ticks or points? Is this backtest or real money trading?

  11. Markus – “You mean 0.4 points, I guess, but now I understand the calculation. The vid is a good idea anyway.” – You’re right it’s 0.4. thanks for finding that, I hate to confuse anyone.

  12. Shodson, sorry but your earlier comment got held up in the queue waiting approval, not sure why.

    I will look at trading spreadsheets. I really enjoy making my own, I like excel. i’m not very good at it but that’s why I like it so that I can learn.

    For expectancy it’s just win % * avg win – loss % * avg loss. Or you can just take the avg P/L of all your trades. Where it gets complicated is if one trades a variable number of contracts. For that I compute the profit of the trade and divide by # of contracts. This weights all trades equally. I’m open to feedback on that. I don’t think van tharp mentions the issue of variable position size. He probably never had that problem .

  13. I looked at trading spreadsheets. Looks nice. Has a lot of things I may not want to track. I don’t see that it tracks MAE which I plan to add to mine. If I could try it before buying it I would like to try it but otherwise I’ll make do with mine. I like how I can just copy/paste trades from Velocity, Ninjatrader, IB, & X_Trader and any other format I want to make. That makes it really easy to enter trades.

  14. Check out my journal post about my spreadsheet.

    Expecting cunparis

    I haven’t incorporated commissions yet but I will. MAE and MFE would be good to add too. I’m pretty good with Excel, so let me know if you need any help.

 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>