Interesting point, very important to understand for successful trading. Let me make it more interesting & complex by sharing some of my backtesting conclusions (I may be wrong..comments welcome..
.
1. Drawdown is directly proportional to Risk. I.e Higher Risk = Higer Return = Higher Drawdown and Lower risk = Lower Return = Lower Drawdown.
2. Drawdown is directly proportional to System Expectancy. I.e Higer System Expectancy = Higher Drawdown and vice versa
3. System Expectancy is inversely proportional to Trade Signal Frequency. I.e Higher System Expectancy = Lower Trade Signals
4. Drawdown Duration is inversely proportional to Trade Signal Frequency. I.e Higher Trade signals = Lower Drawdown Duration.
Now the challenge is to develop a robust system which gives smaller DD and at the same time balances the above 4 points...so how robust is your system..?
BTW, as per my analysis the quickest and easiest way to compound/increase account size and reduce drawdown is by using dynamic risk allocation (risk based position sizing) and regular risk based portfolio rebalance (Only if you have a portfolio) . I.e increase the risk as your account size increases and reduce risk as account size reduces during drawdown.
ncube,
Your post reminded me of my high school math teacher ...with all the 'directly/inversely' proportional pointers
Yes - basically, winning percentage and Reward:Risk decides the system DD w.r.t money and trade signal freq determine the time taken to come out of it. Higher the winning percentage + lower RR or lower winning percentage + Higher RR seems to be the combination. But, most of us, keep trying to get higher winning percentage + higher RR and the elusive search continues. We cant have the cake and eat it too, right ?
Having said all these things, i brought up the lower DD/compounding pointer to drive a point home. General parameter consensus falls along these lines - for intraday system, max DD : yearly return should be atleast 1:7 or something like (1:8, 1:9 or 1:10). For swing/positional systems, it should be atleast 1:5. For example, if one is doing intra in NF, if the max DD of their system is 200, atleast 1400 points should be the average yearly points. Please dont kill me for quoting this measly yearly return points. But, let me assure you a fact- the above-mentioned DD/return numbers can set you free in 3-4 years. We dont need 200-300 points in NF every month on an average. Even 100 points with this kind of low DD is more than what we can eat. Icing on the cake is that this kind of DD can be easily recoverable and also paves a nice way for compounding the account. So, next time when you see a new idea/system, think in those lines.
There are two ways of increasing account size - bringing in more money when you win (but most traders bring in more money after they lose) or compound the account(conservatively or aggressively). In my opinion, once a trader gets past the usual problems of trading, his energy should be largely focused on how to move onto the next level w.r.t trading size. After all, 20 percent return on 10 lacs is much different than 20 pct return on 5 crores.
1) Multiple systems should increase the average daily/monthly returns with same investment.
2) It may increase max drawdown point. A mix of some systems may actually reduce it as well
3) It should decrease the underwater time or turnaround to profit time.
A cumulative analysis of backtest of multiple systems would also be good fun.
lemondew,
Thanks for sharing your view
1. Not necessarily. It will just smooth out the equity curve if underlying principle of the systems are tangentially different or systems trade non-correlated instruments. Actually, it might reduce the average return. On the other hand, the anomaly months(outliers) will be reduced.
2. Yes - both are possible
3. Quite possible - but again depends on the underlying principle of the system or the instruments traded. Almost, all the hedge funds employ multiple systems(sometimes, more than 10) to smooth out the equity curve and improve Risk adjusted returns.
If you get to do cumulative analysis of multiple systems, please post the results here. It would be a great learning for us !!