Rule of thumb for position sizing

G

Ghost of Cutten

If the win rate is low, risk 0.5-1%.
If the win rate is roughly evens, risk 1-2%
If the win rate is high, risk 2-4%

Any improvements on this, critiques etc?
 
By "win rate", do you mean "% winners" (i.e. winning trades as a percentage of all trades)?

If so, then (IMO) the rule misses a consideration of average winner from each system (and profit factor), which ought to be part of your ranking of alternative systems. No? You could have a very high win rate system with a dismal profit factor...

See http://www.istockanalyst.com/financ...t-the-win-rate-profit-factor-and-payoff-ratio

Starting with equation (3) of the above link, and taking r = "avg winner"/"avg loser", you could re-arrange to get ...

avg loser (i.e. what to risk each trade?) = (win rate x avg winner)/[ProfitFactor x (1-win rate)]

[Sorry .... had to re-edit this post, as made several algebraic errors first time .... ]
 
Quote from abattia:

By "win rate", do you mean "% winners" (i.e. winning trades as a percentage of all trades)?

If so, then (IMO) the rule misses a consideration of average winner from each system (and profit factor), which ought to be part of your ranking of alternative systems. No? You could have a very high win rate system with a dismal profit factor...

See http://www.istockanalyst.com/financ...t-the-win-rate-profit-factor-and-payoff-ratio

Starting with equation (3) of the above link, and taking r = "avg winner"/"avg loser", you could re-arrange to get ...

avg loser (i.e. what to risk each trade?) = (win rate x avg winner)/[ProfitFactor x (1-win rate)]

[Sorry .... had to re-edit this post, as made several algebraic errors first time .... ]
More...

Thanks for the link.

Yes, I mean % winners.

I'm deliberately ignoring the profit factor/payout ratio for several reasons. First, for the sake of simplicity. Second, because it's usually impossible to estimate. Lastly, drawdowns are determined by streaks of losing trades, so it is the size of the losses and their frequency that matters most - both of which my rule of thumb incorporates, by sizing to the worst case. Incorporating profit factor could only ever lead you to increase risk (and thus have larger drawdowns), not decrease it. I think this goes against sound risk management principles, namely Murphy's Law and preparing for the worst case.

For example, let's take two systems with high win rates of 75%. One of them has a payout ratio of 1:1 (wins and losses equal on average), the other of 10:1 (average winner 10 times average loser). Now, assume you have a drawdown of 5 consecutive losers (a 1 in 1000 chance). Assuming they trade the same size, both systems will have identical drawdown. The payout ratio and profit factor have no influence at all on the size of the drawdown, when it is comprised of consecutive losing trades - the worst case is identical for both systems. So, I think ditching the payout ratio aids simplicity, renders moot the difficult of calculating it (win rate is normally much easier to estimate than payout ratio), and doesn't affect the viability of the rule of thumb for sizing positions.

Now, the case where the payout matters is if you have say 4 consecutive losers, 1 winner, then another 4 losers, 1 winner etc. Then, having a high payout ratio saves you by clawing back a lot of the drawdown. But assuming a low payout ratio is more conservative. The worst drawdowns will be the consecutive losing streaks. Therefore, by sizing on the assumption of the worst likely losing streak, you are much more likely to get the desired result (avoiding a drawdown in excess of your risk tolerance) than if you try to optimise by assuming a higher payout ratio and some winners to offset streaks of losers.

Does that make sense? Remember, I am not trying to rank systems. I am trading to manage drawdown risk. Whether my system makes 10% per annum or 40% per annum, I still want my maximum drawdown to stay within a certain downside limit. So, how much the system makes is not really that important, what matters is how much it loses during bad streaks.
 
Are you optimizing instrument volatility? 4%, rounded down by an 1/8 of a percent, intraday, works well for SPUS . The Dow is closer to 6% quarter to quarter. And the 4% model began from DOW volatility. FWIW.

So I guess it depends on your style of trading. A good rule of thumb is 1% or 2% of the index, or nominal discount rate of the bid and ask.

The gradation fits. But really where's the advantage if you're risking better than 1:1 against factoring commission, fees, premium on iceberg/OCO, slippage, time, reentry if stopped &c.?
 
Quote from zjulian104:

Are you optimizing instrument volatility? 4%, rounded down by an 1/8 of a percent, intraday, works well for SPUS . The Dow is closer to 6% quarter to quarter. And the 4% model began from DOW volatility. FWIW.

So I guess it depends on your style of trading. A good rule of thumb is 1% or 2% of the index, or nominal discount rate of the bid and ask.

The gradation fits. But really where's the advantage if you're risking better than 1:1 against factoring commission, fees, premium on iceberg/OCO, slippage, time, reentry if stopped &c.?
More...

The volatility of the instrument doesn't matter, because that is already incorporated into your position size. I.e. in a volatile instrument you will trade smaller, in a low volatility instrument you trade bigger, with the result that the % of capital at risk is identical in both cases (assuming win rate is the same).

I'm discussing general position sizing for most/all strategies, not one for a single trading method (in your case, intraday US stock index trading).
 
Ghost, I think you need more than just the win rate to get a
handle on the distribution of drawdowns. That's because most
drawdowns are made up of several alternating loosing and
winning streaks.
Some rules of thumb for the expected maximum drawdown:
- if you have a positive expectation strategy, it will increase
with the log(number of trades)
- if you have a negative expectation strategy, it will increase
linearly with the number of trades
- after enough trades (law of large numbers applies) the shape
of the distribution of the maximum drawdown is determined mainly
by the average and standard deviation of the individual trades.
Higher order moments of the individual trade return distribution
like skew, kurtosis etc, have negligible effects
on the maximum drawdown distribution.
 
Quote from Ghost of Cutten:

If the win rate is low, risk 0.5-1%.
If the win rate is roughly evens, risk 1-2%
If the win rate is high, risk 2-4%

Any improvements on this, critiques etc?
More...

Serious?....Either you are using win rate terminology incorrectly or you have a system that is SERIOUSLY FLAWED!! Any system that has a win rate of 4% means you only win 4 out of 100 trades,...so you LOSE 96% of the time. Flip a coin and you will have a better system. Yes I realize that win rate does NOT effect profitability but it does effect drawdown which means that this system is doomed for failure. This is much like a lottery system. Very low win rate but when it wins I assume you win big. However, as you know most people lose money playing the lottery....

So as for improvements to your system,....dont use it!...Flip a coin and have good money management and risk management. Otherwise you need to risk extremely small amounts. Even 0.5% risk will destroy your account. Especially when you account for the strings of losers you will have with 96% LOSS rate.
 
Quote from N54_Fan:

Serious?....Either you are using win rate terminology incorrectly or you have a system that is SERIOUSLY FLAWED!! Any system that has a win rate of 4% means you only win 4 out of 100 trades,...so you LOSE 96% of the time. Flip a coin and you will have a better system. Yes I realize that win rate does NOT effect profitability but it does effect drawdown which means that this system is doomed for failure. This is much like a lottery system. Very low win rate but when it wins I assume you win big. However, as you know most people lose money playing the lottery....

So as for improvements to your system,....dont use it!...Flip a coin and have good money management and risk management. Otherwise you need to risk extremely small amounts. Even 0.5% risk will destroy your account. Especially when you account for the strings of losers you will have with 96% LOSS rate.
More...

I think a reread of the OP is required.
 
Quote from Ghost of Cutten:

If the win rate is low, risk 0.5-1%.
If the win rate is roughly evens, risk 1-2%
If the win rate is high, risk 2-4%

Any improvements on this, critiques etc?
More...
========
G Cutten
That sounds like a fairly good guideline;
dont we wish we followed that, BEFORE we learned to trade.LOL

4% maybe even lower risk;
if the business is private co or private corp.By that i mean who is the USA ,sells a business if Pakistan threaten to nuke India. ..With a public business plenty sell on 5 minitute chart;
private business probably not even on the radar:D

Of course in a stong public bull market, ignore that ;
in a beartrend -look ouit below-200 dma
:cool:
 
That's not bad.. but check my journal.

Instead of thinking of win rate, you need to think of the probability of the stop hit. I've a system that risks 15% per trade but the stop was only hit like 2x in the last 15 years of trading.

You can't think about risk/reward without thinking about probability.

Also, "risk %" what happens when that level is triggered? Is that per trade? per day? per month? What's the logic behind it?

My rule of thumb is that I want to sustain approximately 12 to 13 losers in a row. However, that doesn't mean I'd have 13 max losses!!

The leverage one can use if they have trading skill and are willing to suffer a 30% or so drawdown is actually way higher then you read about in books.

I could probably risk 8% to 10% per trade and have a fair probability of keeping my drawdown below 35%.

As a rule of thumb, 3%-5% risk per day/trade is moderately aggressive, 5% to 10% is very aggressive.

You need to think about:

Probability of stop hit
and Probability of multiple stop hits
 
You must log in or register to reply here.
Back
Top