1. Instead of "cut your losses early and let your winners run" a trader usually is inclined to "take the profit early and let the losers run". This type of behavior is dictated by emotions while statistical analysis showed that taking profit early leads to a negative expectation. "Let the losers run and averaging losers" in day trading is a recipe of "how to lose your deposit quick".
2. The second reason has to deal with the mathematical fact that for a trader who has a limited capital having a strategy with a positive expectation is a necessary condition to win but it is not a sufficient one. Proper position sizing is required to be a profitable trader.
Let us describe the problem in simple terms. As I posted earlier, day trading can be approximated by a toss of a biased coin. Now imagine that a trader has $1000 and knows a strategy equal to a coin biased as follows - 2/3 heads and 1/3 tails. If this trader every time goes all in on heads than, at some point, all wins and the initial $1000 will be lost. There is a limit to the fraction of the capital this trader can bet on a single toss to avoid the ruin and there is the optimal bet which allows growing the capital with the fastest rate. The mathematical solution to this problem is known as the Kelly criterion. For the coin used in the example above, the Kelly gives 2/3-1/3=1/3 as the optimal fraction of the capital to bet. This solution involves a mathematical idealization that the capital can be dived as many times as it requires and an infinitesimally small bet is possible. Unfortunately, unless a trader has enough money to start with say 100 emini contracts, the Kelly criterion can't be used directly by a small-scale day trader. However, this criterion still can be used to prevent a day trading strategy from running into a ruin.
To summarize, the problem of the optimal bet in the leveraged trading on time frames with the quasi-normal distribution of the return (day or shorter time frames) is still, at least to me, an open question.
I will try to look into this problem.
Update: Leveraged Trading Using The Kelly Criterion.
3. The third reason is the cost of trading which includes trading fees and a slippage. The quasi-normal distribution of the daily return results in close to zero expectation of trading. That is many wins and losses eventually just cancel each other while, as time passes by, the cost of trading steadily adds to the loss. For this reason, many quant strategies that look good on paper do not deliver in real life. Here one has to :
a) daytrade using only liquid trading instruments;
b) choose a broker with a better fee structure;
c) use the realistic cost of trading in calculating the expected return of a trading model.
Let us describe the problem in simple terms. As I posted earlier, day trading can be approximated by a toss of a biased coin. Now imagine that a trader has $1000 and knows a strategy equal to a coin biased as follows - 2/3 heads and 1/3 tails. If this trader every time goes all in on heads than, at some point, all wins and the initial $1000 will be lost. There is a limit to the fraction of the capital this trader can bet on a single toss to avoid the ruin and there is the optimal bet which allows growing the capital with the fastest rate. The mathematical solution to this problem is known as the Kelly criterion. For the coin used in the example above, the Kelly gives 2/3-1/3=1/3 as the optimal fraction of the capital to bet. This solution involves a mathematical idealization that the capital can be dived as many times as it requires and an infinitesimally small bet is possible. Unfortunately, unless a trader has enough money to start with say 100 emini contracts, the Kelly criterion can't be used directly by a small-scale day trader. However, this criterion still can be used to prevent a day trading strategy from running into a ruin.
To summarize, the problem of the optimal bet in the leveraged trading on time frames with the quasi-normal distribution of the return (day or shorter time frames) is still, at least to me, an open question.
I will try to look into this problem.
Update: Leveraged Trading Using The Kelly Criterion.
3. The third reason is the cost of trading which includes trading fees and a slippage. The quasi-normal distribution of the daily return results in close to zero expectation of trading. That is many wins and losses eventually just cancel each other while, as time passes by, the cost of trading steadily adds to the loss. For this reason, many quant strategies that look good on paper do not deliver in real life. Here one has to :
a) daytrade using only liquid trading instruments;
b) choose a broker with a better fee structure;
c) use the realistic cost of trading in calculating the expected return of a trading model.
No comments:
Post a Comment