Well it depends on what and how you trade. I trade a short term (intraday) mean reversion system on only the most liquid securities on the nasdaq and the NYSE. 2007 and 2008 were bumper years for me. 2005-2006 and 2009-2010 were more "normal". But that's the way the system was designed. A bit like Universa except I actually make money most years lol. And I agree you shouldn't compare to "pro hedge funds". Its alot harder to generate substantial ROI when you have billions as opposed to hundreds of thousands. Unless of course you are Renaissance. Lol.
Being new to understanding Kelly % (but not new to trading), I too ask this same question. Although I could risk $38k, in reality I'd only risk about 1% of capital per trade. I would never let it a loss go to $38k unless it gapped way, way down. which would be a crash in my opinion.
The only clarification I'd offer is that if he's daytrading, a stop loss is OK. If he's holding overnight, he should be using options. A stop loss will not protect you against an overnight gap that goes right through it and beyond.
The 38% would be the equivalent of "value at risk" so that if your initial stop loss as hit, you would take a 38% loss. But, let's say you move your stop to cut your risk in half and it gets hit, your loss would be 19%.
Yes I do use options not stops. Problem arises when for example Lumber doesn't have an equivalent ETF and corresponding options. I know there are options on futures but they are too expensive with huge spreads and low volume. So anyway the Kelly % should be the amount I've bet - but not the amount I risk (due to use of options to limit losses to about 1-2%)? is this right?
Should be the amount you bet to get the maximum growth based on your winning percentage and the amount you win or lose on each trade. So yeah, you could do 38k if you hedge it correctly, according to Kelly. In real life, you might not want to do the max according to Kelly, unless what you're using is money you know you could live without. My risk capital is a small percent of my total, so I can use Kelly with abandon, knowing I'm not going to have to eat cat food if I screw up big time. Your mileage may vary.
Thanks so much for clarifying this. Your explanation is what I thought Kelly % meant but I wasn't sure. I like that you metioned something I think is important: to use money you know you could live without. So if your trading capital is a small percentage of the total money you can live without, you can risk the full Kelly % all day (or night as the case may be) long.
OKAY! I found the answer myself. Some of you suggested monte carlo, others suggested that there is NO normal distribution, and others said trading results are normally distributed but is proprietary info from the brokerages. Well I will share with all of you generous people the results of my findings (and intuitions): I was thinking about it (in the bathroom no less), that futures trading results HAS to be normally distributed around some % return be it -1% or 0 or +2% over some period of time. Here is the answer. The study, over 15 years of data from Barclays CTA Agriculture index return, shows the normal distribution of results from professionally managed futures accounts is: 0.40% / month Normal distribution of trading results If I'm getting 2% / month over 3-4 months on a 100k account, then I'm doing WELL. Sure my trading period is not enough time to make a very strong argument that I have skill and not just luck, HOWEVER, bottom line is my account is increasing in value by 7.5% over 3-4 months. In addition these past summer months were considered "difficult trading" by most professionals. By the way, I practiced for 1 month papertrading before going live on May 25. I was short the flash crash and made a 10%+ return ($32k on a $300k account). One could add these results too - but I don't because it wasn't real money. Anyway, now that I've determined the answer myself and learned a lot because of you all: I just wanted to say thanks again for taking the time to teach. I hope this last post has helped you all learn too about the normal distribution of futures trading (my original question). With respect to my trading colleagues, Anesthesiaman
OKAY! I found the answer myself. Some of you suggested monte carlo, others suggested that there is NO normal distribution, and others said trading results are normally distributed but is proprietary info from the brokerages. Well I will share with all of you generous people the results of my findings (and intuitions): I was thinking about it (in the bathroom no less), that futures trading results HAS to be normally distributed around some % return be it -1% or 0 or +2% over some period of time. Depends on how strict you want to be to a truely normal distribution. The main issue is kurtosis, the value which describes the 'look' of the distribution. Market data usually is "Leptokurtic" meaning it has an over abundance of data about the mean (generally 0, sometimes a bit above 0 = skew) and the over abundance of fat tails (data that falls far from the mean). A true normal distribution would be more 'squat' that is less near 0 values, more values in the middle (say between 1-3%) and far less over 5% values. So qualitatively, yes the market is 'normal' but quantitatively, it is not. Here Here is the answer. The study, over 15 years of data from Barclays CTA Agriculture index return, shows the normal distribution of results from professionally managed futures accounts is: 0.40% / month Normal distribution of trading results If I'm getting 2% / month over 3-4 months on a 100k account, then I'm doing WELL. Sure my trading period is not enough time to make a very strong argument that I have skill and not just luck, HOWEVER, bottom line is my account is increasing in value by 7.5% over 3-4 months. In addition these past summer months were considered "difficult trading" by most professionals. But were you trading agriculture-related stocks? I believe you are pushing the limits of comparative statistics. Kind of an apples to oranges comparison. By the way, I practiced for 1 month papertrading before going live on May 25. I was short the flash crash and made a 10%+ return ($32k on a $300k account). One could add these results too - but I don't because it wasn't real money. Anyway, now that I've determined the answer myself and learned a lot because of you all: I just wanted to say thanks again for taking the time to teach. I hope this last post has helped you all learn too about the normal distribution of futures trading (my original question). With respect to my trading colleagues, Anesthesiaman As for how I would approach your quandry. So to paraphrase, you want to compare your trading results to chance. Chance redefined is a random entry. Therefore you need to compare your results to random entries. Given you use of ToS (which has the worst paper trading platform ever, by the way), my guess is you cannot really backtest rigorously. If you could, you should just place random entries into the market and see the results compared to your results, looking at variables such as profit, std. dev of returns, profit factor, win%, etc. Choice of variables may be critical...I don't know. Without backtesting, what I would do is roll 3 dice. 1 would be for the hour, 1 for the 10's of minutes, and 1 for the minutes (accepting the fact that only 6 of 10 possible minutes would result). So you roll a 3,2,5. You would place a trade in the third hour of trading at the 25 minute mark. Hold that trade for a period approximating you real trade time, and record the results. It's crude, kludgy, and non-robust, but it will provide some evidence on random market returns. Glad to see Mike and Bill (as always) and good thread so far, hope my suggestions are logical and helpful. Masterjaz
Very insightful and helpful post masterjaz. Yeah I learned about leptokurtosis after reading up on the link that I posted about Barclays CTA's trading returns above. Although the above results seem to be exclusively ag futures, I only trade these on a rare occasion, so there probably are some differences when comparing to what I trade. That being said however, I cannot imagine other pros who trade the other non-ag futures to be fantastically different from the results here (although it could be). Anyway the whole reason why I began trading in April-May 2010 was that my cash was parked in a crappy savings account earning next to nothing. Since I didn't have to study medicine as much anymore (because i just finished residency) and because I "remembered" my trading ability that I never really took seriously b/c of being focused in medicine, I decided to apply what I knew to futures (i used to trade options only) and hope for a better return than a savings account. Glad I did! I like your idea about the dice Masterjaz. I thought that was funny even though it seriously would work since I do wish to compare my results to chance alone. But I also want to compare my results to professional active traders, SP500 return, risk-free treasuries (oxymoronic these days...), LIBOR, etc. Since I have no idea how to backrest my discretionary trading, that may be what I have to do ultimately. In any event, if the pros average return is better than mine, I'd rather give them my money to trade (if I knew who they were). But then I don't want to chase returns either (and don't trust pros look at Gartman's recent interview yesterday with Bloomberg), and so I'm back to square one - trading for my own account - and trying to determine whther continuing to trade is a worthy endeavor given my results. Hence the creation on this thread. But intuitively I already know all the answers (and yet know nothing at all...): I have some trading skill that can be developed further to improve results by hedging less, not selling so soon to capture more profit and start trading up to my Kelly %. If I didnt know wht i was doing, that sounds a little like a recipe for an account blow-up. But I'd only do it if I could trade full time and really monitor my positions rather than part time, end of day as I do now.