Systematic Equity Investment Portfolio Performance Tracking

[XRAY27]

Few reasons, though hedging looks good in hindsight, my backtesting had not shown any advantage as these kind of falls are quite rare. Also my strategy has no direct correlation with Nifty most of the time, hence Nifty derivatives cannot be used.

In the backtesting, my strategy has already considered 2008 fall...with max DD of 30% so this time if the fall goes more than 30% it will be in untested waters..

Actually my portfolio allocation system has already given a trigger to reduce equity exposure..but I am delaying it as I am comfortable with current DD levels and anyway I plan to do tax loss harvesting this month...will rebalance my portfolio beginning next month if required.

Respect risk and system !!

 

[ncube]

Respect risk and system !!

Thanks, yes I am following the system not over riding it..looks like I did not put it correctly in my post. As I run multiple strategies I have a signal at portfolio allocation level which will indicate the allocation to be done for each strategy. That signal was already triggered last week and I have been reducing the equity exposure as per the signals generated in each strategy.

For this particular strategy that I am tracking here the signal is not triggered at strategy level. It has built in exit rules with will trigger automatically...which I am expecting to happen once the DD crosses 30% as that was the max DD of this strategy and had happened in 2008. If the present trend continues the strategy will automatically start scaling down in coming days.

Though the global portfolio allocation signal has triggered for reducing the equity exposure, I do not want to override the strategy level signal as it has already been stress tested in backtesting and I am comfortable with the current risk.

Knowing when to break the rule is very important but difficult to make as I run a fully systematic and mechanical system..

 

[TracerBullet]

For this particular strategy that I am tracking here the signal is not triggered at strategy level. It has built in exit rules with will trigger automatically...which I am expecting to happen once the DD crosses 30% as that was the max DD of this strategy and had happened in 2008. If the present trend continues the strategy will automatically start scaling down in coming days.

Not saying its wrong and my experience with longer term systems is none. And i am assuming things.

But here goes - how do you get confidence in a rule that seems to be based on a single event (2008 bear market). In general the sample size for most rules and conditions would be very low in longer term systems - how can you be sure that any rule is significant and contributes to the edge without becoming an over-optimization.

 

[ncube]

Not saying its wrong and my experience with longer term systems is none. And i am assuming things.

But here goes - how do you get confidence in a rule that seems to be based on a single event (2008 bear market). In general the sample size for most rules and conditions would be very low in longer term systems - how can you be sure that any rule is significant and contributes to the edge without becoming an over-optimization.

Excellent question, answer to this is the real key to designing successful trading strategies..

Typical mistake most traders do when designing a strategy is not knowing why their strategy should work in the first place. They think if the back-test results are good the strategy would perform the same in future which will never be the case. Any strategy we design based on back-testing is nothing but over-fitting/over-optimization.

There are 2 key ways one can design the strategy,
1. Data mining (Back-testing set of rules)
2. Hypothesis/Factor based

Data mining is the most common used by most traders where one just identify few rules and back-test it on historical data and find whether the rules work, in extreme cases one ever fine tune the parameter values.. like if price go above last 6 bars and above 28 day ema buy etc..

I dont encourage data mining/back-test based rules for designing a strategy as it is always based on hindsight bias and no guarantee it will work in future. Instead I design strategies based on core concepts by first defining my hypothesis/factor on why the strategy should work. For example, why Value, Growth, Pair, mean reversion or momentum concepts would work..what factors drive it.

Based on these factors I design my strategy rules even before back-testing it..This gives me confidence as I know why a particular rule will work and should work in future. Once this is done, back testing is done not to fine tune the strategy but just to capture the performance parameter in order to accept or reject the hypothesis.

So to answer your question my strategy is not based on 2008 event or any historical event for that matter, just that I have seen the worst DD for my strategy in 2008 and also I am not sure how far current DD will continue as once it reaches 30+ it will be in untested waters...But I know the reason for this DD and confident on the factors...

Hope I am clear.

 

[XRAY27]

Knowing when to break the rule is very important but difficult to make as I run a fully systematic and mechanical system..

Don't do the rule breaking i,many lost there way in the market (rule breaking way) ,carry on your time tested way..best of luck..investment or trading will have underwater area for some time,but when we follow the method ,we move on to new equity curve ..

 

[XRAY27]

Excellent question, answer to this is the real key to designing successful trading strategies..

Typical mistake most traders do when designing a strategy is not knowing why their strategy should work in the first place. They think if the back-test results are good the strategy would perform the same in future which will never be the case. Any strategy we design based on back-testing is nothing but over-fitting/over-optimization.

There are 2 key ways one can design the strategy,
1. Data mining (Back-testing set of rules)
2. Hypothesis/Factor based

Data mining is the most common used by most traders where one just identify few rules and back-test it on historical data and find whether the rules work, in extreme cases one ever fine tune the parameter values.. like if price go above last 6 bars and above 28 day ema buy etc..

I dont encourage data mining/back-test based rules for designing a strategy as it is always based on hindsight bias and no guarantee it will work in future. Instead I design strategies based on core concepts by first defining my hypothesis/factor on why the strategy should work. For example, why Value, Growth, Pair, mean reversion or momentum concepts would work..what factors drive it.

Based on these factors I design my strategy rules even before back-testing it..This gives me confidence as I know why a particular rule will work and should work in future. Once this is done, back testing is done not to fine tune the strategy but just to capture the performance parameter in order to accept or reject the hypothesis.

So to answer your question my strategy is not based on 2008 event or any historical event for that matter, just that I have seen the worst DD for my strategy in 2008 and also I am not sure how far current DD will continue as once it reaches 30+ it will be in untested waters...But I know the reason for this DD and confident on the factors...

Hope I am clear.

Good points for both trader and investors..

 

[ncube]

Portfolio Status after 13% fall in Nifty today:

 

[ncube]

Date: 28-03-2020

1. Tax loss harvesting done this week, hence allocation to Equity portion is reduced. Will rebalance the portfolio next week based on the overall market trend.
2. The max DD for the strategy in this fall is about -32% and its the highest so far in the last 20 yrs. Currently its about -24% as there was recovery in the last 3 days of the week.
3. The max DD for the strategy at the portfolio level was about -18% considering 30% in Debt Funds,
4. Total return for the strategy for the FY 2019-20 is about 4.2% considering the returns from Debt funds and regular profit booking from the strategy.

Portfolio Status:

 

[ncube]

Date: 31-03-2020

End to a difficult month and FY 2019-20. FY Ended with returns of about 8.1%, slightly better than the FD returns, but has beaten the Equity MF & Nifty Index returns handsomely.

Month ended with a loss of abt -17% at one point it was about -28%...

Going into next month with Equity portion fully invested.

 

[ncube]

Date: 14-04-2020
1. Equity portion is fully invested and is in a comfortable ratio of 70:30 (Equity : Debt)
2. Healthcare sector allocation has grown over 41% in the portfolio followed by materials at distant 16%, will be booking profits in coming days.

Strategy Equity Curve:

Sector Allocation: