Acrary is not 100% reliable
- Thread starter es175
- Start date
Basically that's what I have already done.
Over the past 6 months I worked on building a data/pattern mining framework to see if I can get any statistically significant patterns that may predict (to some extent) a type of the next day. My conclusion is that it is not possible.
The only marginally useful thing I have found is that there are some conditions/patterns that predict the targeted behavior is not likely to be present next day. It reduces number of trades and increases hit rate. But that's not enough to build a consistently profitable model.
Then I moved to intraday price analysis, and I'm still there to test ideas. Nevertheless, I have already built trend-following model that produces better than random (edge test) results over the years (out of sample actually).
Next thing I have tried to build a complementary model for my trend model. A counter-trend or range-bound trading model, but so far no success. It may be I have exhausted my imagination and cannot think of any useful way to code up a counter-trend or range-bound entry. Or it may be that I have never tried to trade these concepts. My previous experience was with trend-following only.
There is no drag with using a % of your capital to size bets. The drag when you have a winner followed by a loser (or vice versa) is offset by the reduced losses when you have a string of losers, and the increased profits if you have a string of winners.
If you use a fixed nominal sum, then you end up trading excessive size after a losing steak, and trading too small after a winning streak. This increases the risk of blowups and reduces the benefits from compounding - both of which have a huge influence on results over the long-term.
Conclusion - unless you want increased blowup risk and a lower average compound annual rate of return, it is better to bet a fixed % of capital on your trades, not a fixed nominal sum.
If you use a fixed nominal sum, then you end up trading excessive size after a losing steak, and trading too small after a winning streak. This increases the risk of blowups and reduces the benefits from compounding - both of which have a huge influence on results over the long-term.
Conclusion - unless you want increased blowup risk and a lower average compound annual rate of return, it is better to bet a fixed % of capital on your trades, not a fixed nominal sum.
How does the optimal position sizing method for a trader have anything to do with what a hypothetical boss thinks of it, or "trading as an engineer" - whatever that means. Why discount the effect of strings of winners? It affects your results and you will get them just as surely as you get strings of losers. And don't forget that the "drag" works in reverse when you get a string of losers. Are you telling me that if your account falls 30%, you are still going to be betting the same size as beforehand?
Betting using fixed nominal sums seems logically inconsistent. At some point you are going to scale up size because your capital has increased (or reduce size if you have losses) - otherwise you would still be betting the same size when your account is up 10 fold. So you basically end up sizing bets as a % of capital anyway, just with an intervening period where you completely arbitrarily vary your position size (and thus risk) as a % of capital. It seems illogical to ignore account size for an arbitrary period, and then all of a sudden take it into account. If account size is a key determinant of the optimal bet size (which it clearly is), then that is true at all times and on all trades. Your past trading results have no bearing whatsoever on what the optimal bet is for the trade you are facing now - the optimal bet size is purely dependent on the risk, reward, win rate of the trade and so on as it stands now. Your past results are irrelevant to the correct bet size at the present time, otherwise you would be advocating that two people with identical capital, identical views on the utlity of money, identical views on risk preference, and facing the same trade, should trade massively different size, purely based on their past results. That doesn't seem to make any sense.
Why size based on prior capital rather than current capital, why trade more size as a % of capital after drawdowns, why trade lower size as a % of capital after a winning streak? Why should identical traders take completely different position sizes when facing identical trades?
Betting a fixed nominal sum seems to have multiple drawbacks and no benefits.
Great post.
That's what I meant when I said you only need to learn one system, or modality of trading ... what that will do is begin to show you how to classify the other types of markets ... but remeber, you do not have to, to succeed as a trader, or to make money.
Quote from OddTrader:
I'm afraid the opposite of the above would be Also true, at least some of the times.
What's your reasoning to back up this claim?
Your conclusion would indicate that two traders, identical in all respects (risk preference, capital, methods etc) would be obliged to bet hugely differing amounts on exactly the same trade in some situations. This is completely irrational.