System Statistics

I'm making all sorts of graphs and spreadsheets to monitor two systems I've created, I'm using the usual sharpe, R-Squared, Win/Loss% and win$:Lose$ figures. I have the standard deviations and all trades, winning trades and losing trades. Thing is, where could I go to read up on what kind of figures/results I should be looking for.

hmm, tricky one, but the one i look for when devising strategies is the time between equity highs.


this is probably the best reflection on the tradeability of a method, as it reflects the psychological problems that occur when trading mechanical methods which can result in several losing/breakeven months.

as long as the drawdown isnt spectacular, then i feel that this is the best guide to system performance.

FC

Agree, about every 3 weeks this goes into a P/L sideways/slightly-down movement. I've found that adjusting the size traded based on volatility and the addition of stops/targets helps a lot, it also means the drawdown tends to just become a sideways movement for a number of days/trades. What I'm wondering is whether I should be trying to push my analysis towards improving the sharpe ratio, the win/loss %, the $win:$loss ratio or should I just leave it alone and only tinker when things look like they're going wrong?
Was just wondering if people record their results like this and if there are certain factors that outweigh others.
Thanks all

ah right. ok. i see where you are coming from..

when it comes to tinkering with strategies, the one i like to employ is improving the win rate. its always easier to trade a 70%+ system, than anything lower.

i once came up with a <10% system win rate. tried trading it for a month. didnt last!!!!

im always happy to forego an overall bit of profit, in terms of increasing the win rate.

again this has to do with psychological reasons..


a tip for improving the win rate of most systems...

remove the bad trades, or even the 50/50 trades.

eg if you have a system that overall has a win rate of 65% , if you then find a common pattern in those trades which only is correct maybe 50% of the time (but might still generate a profit) , then exclude those trades from your overall method. then bingo your strategy leaps up a few %, and saves on transaction costs etc...

another way to increase overall profit is to find a similarly accurate system, but which trades from an uncorrelated set of rules. then disregard trades signalled in opposite directions by both systems.

again this will boost the win rate, and profit per trade...

FC,

hmm, tricky one, but the one i look for when devising strategies is the time between equity highs.

I also have been trying to decide which figure to use as an indicator of a 'good' system. The idea of time between equity highs is excellent.

I had been trying to convince myself that return on drawdown is the factor to follow, but that doesn't take into account the mental side of coping with a string of losses. A high win rate is also a definite must for me.

pogle

I like using the linear regression function in excel graphs, it has a good slope and R-squared is very high and I like the fact the curve is nice and not too bumpy. This makes me feel fairly confident with it, it doesnt have a great win% though, just over 50% which normally I would hate to trade but when I combine it with another strategy I get good day win%, I'm trying to focus on this % over the individual trade% as at the end of the day its the profit you've made at the close of business and not on each individual trade.

Erm, why not spend the time honing/devising your systems instead?

When I first started trading I spent 10 times as much effort on devising graphs and spreadsheets to capture every nuance and the very essence of my sheer trading brilliance...

...before I realised I wasn't making any profits.

Just a thought.

FetteredChinos said:

its always easier to trade a 70%+ system, than anything lower

Click to expand...

FetteredChinos said:

im always happy to forego an overall bit of profit, in terms of increasing the win rate.

Click to expand...

I totally agree with FC. At the end of the day, you have to be able to sit there and trade the thing with the phone ringing and with your wife talking about this, that and the other and without having a panic when you suddenly meet an unexpectedly large drawdown.

Hello

I've had a few months tinkering around with a system that was 55-60% win probability per month using a target with a win ratio of 1.5
I could never quite improve the win prob % until I changed the win ratio to 0.75 which caused the win prob to jump to 70-75%! It's less roi, but I have fewer negative days.
Easier on the old tiker
As FetteredChinos states, it is much easier to trade a system based on a high win%.
I also have set trading hours.

Minx, do you use targets?

Cheers
C

The Bramble: I'm doing all the statistical work to see if there are any ways to increase the efficiency of the trading method but also to try be aware of any overfitting of my results.

cd173: I do use targets for part of the position and the remainder as a runner until I get an exit signal. This is another reason for me collecting and analysing data so I can find max favourable excursions and then the actual close of the position. From all this I've found that certain trades will go onside by quite a bit but also give a lot of the gains up, hence I've plopped in a couple of targets and it has improved the P/L. This kind of trading helps me as once the targets start getting hit I feel so much more relaxed knowing that a sudden move wont hand all my money back to mr.market.....

I dont think concentrating on the win % is necessarily a good idea, its the profit ratio that matters equally. I tinkered around and found a system that gave lovely win percentage but the ratio was rubbish. In order to get a good win percentage the system had used trailing stops which got next to nothing. But it was hard to keep close stop losses and they woud invariably get hit 1/4 of the time and wipe out the last 6 trades profit. Rubbish. However, some systems that try and fail and try again to get in at the perfect point often succeed to get nice moves, the profit being many times that of the losing.

One attribute of a system that I think is very important to keep in mind is open draw down! Some trade software doesn't include this but its a bugger! I was enticed into the heady world of Metastocks add-ons (a huge con), I tested out the profitability of the systems and thought how incredible that they are giving away a system that has 80%wins. Until I looked at the small details, the trades didn't use stops and the open positions would sometimes suffer such huge draw downs before it close higher, it would have wiped you out! Or you would have got cold feet and bailed with big losses before it got a chance to change direction. But on the closed trades chart the drawdown looked acceptable.

Its always good to know when constructing a system how its going to behave during open trades. Its also good info to know when calculating stop losses and position sizing. If you have an 80% winning system but it often draws down before up, it might be worth trading but with smaller size. But if you saw that 80% and thought it was easy money and could bet your whole wad on it, you'd be biting the dust pretty quick.

Be wary of just looking at the winning percentage.