Australian (ASX) Stock Market Forum

Difference in Win % : Short-Term vs. Longer-Term Systematic Trading

Joined
Apr 14, 2007
Posts
317
Reactions
0
Hello All,

It seems a popular belief that if you look at the stats of the majority of profitable short-term systems the Reward:Risk ratios are normally low (1:1, 2:1, etc) whilst the win rate percentages are normally high (70%+ from what I've read). In constrast, profitable longer-term systems generally seem to exhibit the opposite; higher Reward:Risk ratios but with lower win % (anything around 40%-50% doesn't seem uncommon).

Although I haven't spent any time developing short-term systems, my own experience with longer-term systems seems to echo this view.

Consequently, my question is why is this normally the case, and more importantly why can't a longer term system also generate a high win rate %?

To expand this further I'd like to focus on what contributes towards a higher Win Rate %. IMO Possible contributers could be:

1. The reliabilty of the actual entry signal must be very good to begin with. By reliable I mean that the possibility of the trade moving in your favoured direction AT LEAST past the BREAK-EVEN POINT must be v.high.

2. A target stop is maybe being employed which would limit the R:R. However, given that the only 2 exit outcomes are either (i) Target stop being hit (resulting in a Win) or (ii) Initial Stop being activated (resulting in a Loss), then point 1 (reliability of entry signal) still needs to be met to achieve a high win rate, therefore a target-stop probably doesn't contribute that much towards a higher win rate.

3. Using a Break-Even stop, which if activated would result in more "slight profit" or simply "scratch" trades as opposed to actual losing trades.

4. Using a tighter trailing stop, which would result in a shorter holding period. However, if one is utilising a break-even stop then any actual exit point above this break-even stop would result in some form of profit and thus contribute towards the higher win % anyway, so like the target stop, a tighter trailing stop probably doesn't contribute that much towards it.

As a result, it seems that the key points to increasing the win rate % is:

1. The realibility of the entry signal.
2. Adopting a break-even stop.

If this is the case, then surely a high win rate can be achieved with longer-term systems by focusing on these 2 points?? Point 2 would be easy to adopt so the reliance on achieving a higher win % is with the reliability of the entry. Howvever, as most longer-term systems have a lower win rate%, is the reason for this that finding a reliable entry signal is more difficult?
I ask as in my own situation, regardless of what approach I adopt for my entry and exit rules, my win rate % never seems to exceed 50% and in most cases falls within the generalised 40-50% range.

I'm sure I've missed some very obvious points and welcome any comments / discusion on this.....


Thanks in advance,
Chorlton

PS. I fully understand that the "expectancy" of a system is what really matters in terms of making a profit or not but making a profit in not really the point of this particular discussion.
 

Wysiwyg

Everyone wants money
Joined
Aug 8, 2006
Posts
8,424
Reactions
262
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hello All,


Consequently, my question is why is this normally the case, and more importantly why can't a longer term system also generate a high win rate %?
As a matter of fact I am testing rule based auto -traders at present over an extreme 5 month peak - trough - peak cycle and all except a few start out great but overtime run negative.This is with human programmed strategies so it would be a likeness to manual strategy trading.

I can only think that it`s longer exposure to market gyrations increases the likelihood of a major drawdown as we have all seen over the past year or so..Short to medium term seems statistically more advantageous.

:2twocents
 
Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Entry signals for longer-term systems can't account for technical and fundamental changes further along time-wise. I think it wouldn't be hard to devise an entry signal for a longer-term system that is highly reliable after say one or two days but that reliability drops off over time to around the 50% mark. But every day beyond entry increases the probability of an event that will impact reliability. This at least has been my experience when testing systems.

I did create a long-term system that had a win rate above 90% but it was untradeable due to the open equity drawdown. Closed equity grew beautifully but who wants to sit thru massive open equity swings?
 
Joined
Apr 14, 2007
Posts
317
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Wysiwyg,

Thanks for your thoughts....


MS+TradeSim,

Again, thanks for your comments.

I think it wouldn't be hard to devise an entry signal for a longer-term system that is highly reliable after say one or two days but that reliability drops off over time to around the 50% mark.
Surely once the trade has moved past "Break-Even" which (assuming the trade is initially favourable) should occur within a short-period (1-2 bars) , then as long as a Break-Even Stop is being employed the trade should result in some form of profit or at the very least become a scratch trade? Either way the result would be that the win rate would be higher?

Clearly from my own testing this is not the case, which would suggest that the Intial Stop is being hit first, which in turn would suggest a less reliable Entry Point to begin with.
 
Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hey Chorlton,

Your line of thought makes sense but I don't use breakeven stops on longer term trades. I only use a trailing stop that is volatility based and is also used for position-sizing. The testing I've done tends to indicate signals which have a high probability of going up short term don't translate reliably into trend entries. I suspect this flows from a common situation of momentum driven moves collapsing quickly rather than continuing as new trends. Needless to say, one might find different results by approaching the problem with different logic.
 
Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Just ran some tests out of interest. I took my trend system and used time stops without initial or trailing stops.

Exit after X days - Win rate - Ave.MaxDD - Ave. Return (Jul.00-Jan.09)
1.....48%...19%...1970%
2.....50%...21%...2362%
10...65%...14%...3690%
20...61%...17%...5225%
30...64%...32%...2793%

On first glance, exiting after 20 days seems great but many of the winning trades in there have MAEs above -20% and some are above -50%. They were possibly only saved by the roaring bull market.

In comparison using the trailing stop (based on my own formula), the same entry signals hold winners for an ave 44 days, losers 19 days, average win rate 45%, ave MaxDD 19% and av. return over the same period of 7000%. The majority of winners have MAE of less than -20%.

I have not yet tested how these results would be affected by using breakeven stops but I am initially thinking that win rate might increase however profitability would drop. On the other hand looking for a highly reliable entry for a long term system may result in so few trades that the overall return diminishes.
 

tech/a

No Ordinary Duck
Joined
Oct 14, 2004
Posts
19,447
Reactions
3,701
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Gents.

The answer is the trend.
Longterm you wont be trading anywhere near the number of trades as a short term system.
Most long term systems will see you in a trade/trend for 12 mths or even more.
If you take 10 positions it may take 30 trades before your portfolio is full and your only taking trades when your exited or stopped.
Hence the 33% or there abouts win rate.

Your R/R will be much higher.

Short term you'll have many more trades and a great deal more in win rates due to the number of trades and most importantly the number of trends.

If you could take EVERY trade in a longterm system each time its generated you'll find the win rate is highly correlated to the length of the trend.

If you have a system that has an average hold for 12 mths and the trend goes for 5 yrs then the win rate will be high.

Thats why really short term systems have very high win rates.
You get many many more 1-4 day trends than months or years of trend.

When you think about it its not the entry or exit but the RIDE!--trend.

Now when I found this it was like a blinding flash of the obvious!
 
Joined
Apr 14, 2007
Posts
317
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi MS+TradeSim,

I have not yet tested how these results would be affected by using breakeven stops but I am initially thinking that win rate might increase however profitability would drop.
Maybe I'm incorrect in my thought process but just because the MAE of some of your trades were above -20% this does not necessary mean that you would be stopped out more by using a Break-Even (BE) Stop. In my mind it all depends at which point in the trade the MAE occurred. If for example it occurred ABOVE the BE Level then profitability would not be affected but win rate % would increase.

In my mind, the key is how quickly a trade exceeds the the BE level.

On the other hand looking for a highly reliable entry for a long term system may result in so few trades that the overall return diminishes.
Yes this a good point. However, if fewer (but more reliable) trades also resulted in a much lower MaxDD, one could implement leverage to increase those returns? Just a thought....


On a side note, have you carred out any backtesting of your system/s prior to 2000 ?? If so, I would be interested in your "yearly" return for the year 1999 and more importantly how that year compared to others.
From testing of my own long-term systems (testing between 1998-2004) I have found that all my systems perform ridiculously well in this particular year. Returns are around 3-4x more than any other year tested. Consequently, I'm trying to understand why this is happening.

Regards,

Chorlton
 
Joined
Apr 14, 2007
Posts
317
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi Tech,

As always, You make some interesting points.

Longterm you wont be trading anywhere near the number of trades as a short term system.
Most long term systems will see you in a trade/trend for 12 mths or even more.
Looking at the results of my own Long-Term systems, the Avg Holding period for Wins is 6mths & Losses is 1 month.

Given my Capital Base, Max number of positions is around 10-12 and as it currently stands I have 2-3x more signals than actual trades, which is due to lack of available capital.

Your R/R will be much higher.
Agreed. Looking at my own stats this is definately the case

Thats why really short term systems have very high win rates.
You get many many more 1-4 day trends than months or years of trend.
Is this really the reason? I agree with the 2nd line in that you will get fewer more established trends compared to the 1-5 day trends but IMO this doesn't neccessary equate to a better win rate.

As long as the trend lasts for as long as the trade needs to reach its Break-Even point then even if it is then stopped out (due to the trend terminating) the result should still be a similar win rate to that of a shorter-term system? Or am I missing something,( which is probably likely!!) ?
 

tech/a

No Ordinary Duck
Joined
Oct 14, 2004
Posts
19,447
Reactions
3,701
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

As an exercise.

Grab a chart say BHP
In the direction your trading and I suggest going with the prevailing trend.

Using your favorite entry check the following. (Check each entry even if its in a trade already held in a timeframe)
How often do you get a 1, or,2 or,5 or 15,or 150 bar trend following your signal and B/E.

Havent done it for a while but I' reckon you'll find the amount deteriorates (Number of trends) the longer you go.
 
Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Maybe I'm incorrect in my thought process but just because the MAE of some of your trades were above -20% this does not necessary mean that you would be stopped out more by using a Break-Even (BE) Stop. In my mind it all depends at which point in the trade the MAE occurred. If for example it occurred ABOVE the BE Level then profitability would not be affected but win rate % would increase.

In my mind, the key is how quickly a trade exceeds the the BE level.
I understand where you're coming from. It would be interesting to check.

Yes this a good point. However, if fewer (but more reliable) trades also resulted in a much lower MaxDD, one could implement leverage to increase those returns? Just a thought....
Certainly over long-term trend following that would be probable and something you should test for.

On a side note, have you carred out any backtesting of your system/s prior to 2000 ?? If so, I would be interested in your "yearly" return for the year 1999 and more importantly how that year compared to others.
From testing of my own long-term systems (testing between 1998-2004) I have found that all my systems perform ridiculously well in this particular year. Returns are around 3-4x more than any other year tested. Consequently, I'm trying to understand why this is happening.
My systems all use a filter that does not extend back prior to mid 2000. If I take out that filter so I can test earlier, 1999 would be very good but skewed upwards due to 2 trades. Even if I take out those trades the year would still be in the 300-400% range. When I check forwards without the filter, results are significantly degraded over all. It appears 99 probably would have been a very good year. It was the end of the dot com boom so perhaps systems were just collecting some of the huge runs.

Just for interest's sake to illustrate what I'm saying above, here are 3 charts. The first is annual return without the filter. As you can see 1999 would have been the best year. But given the annual variance I would not even trade the system.

Chart 2 is how the unfiltered system would travel from July 00 - Jan 09.
Chart 3 is how the filtered system travels over the same period. (I stopped trading it live in Dec.07 or Jan 08 (I forget off-hand) and am still waiting for the sewers to unclog before it goes live again.) I do have a short-term system which is currently live and tracking well.
 

Attachments

Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi Chorlton, I'm going off-track from your OP but I will return to track after I test some breakeven ideas.
 
Joined
Apr 14, 2007
Posts
317
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

My systems all use a filter that does not extend back prior to mid 2000. If I take out that filter so I can test earlier, 1999 would be very good but skewed upwards due to 2 trades. Even if I take out those trades the year would still be in the 300-400% range. When I check forwards without the filter, results are significantly degraded over all. It appears 99 probably would have been a very good year. It was the end of the dot com boom so perhaps systems were just collecting some of the huge runs.
Hi MS+TradeSim,

Thanks for taking the time to reply. It's interesting to see that your system also achieves a v.high return over the same year. Its just reinforces the fact that 1999 was obviously an excellent year for long-term systems.

The returns of your filter system are IMO very impressive. Can I ask what kind of MaxDD was exhibited during this same period? In addition, may I ask what your parameter criteria was during this testing? eg Starting Capital, Risk per Trade and whether profits were compounded?

As an observation, looking at your own charts (and based on my own developing experience to date) is the importance on having a good year to begin trading ones system as the returns in subsequent years can vary dramatically depending on how profitable the initial trading period is!!
I think this is due to the amount of initial capital one uses to fund the account. Unfortunately, one cannot predict what kind of year lies ahead in advance!!!

Out of interest, (if I may ask) why does your filter exclude ALL trading activity on the ASX prior to 2000? Have you actually chosen not to include those prior years (for example as a result of poor available data) or is it just a coincedence based on your filter condition?

Just interested...

Kind Regards,

Chorlton
 

tech/a

No Ordinary Duck
Joined
Oct 14, 2004
Posts
19,447
Reactions
3,701
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi MS+TradeSim,

Thanks for taking the time to reply. It's interesting to see that your system also achieves a v.high return over the same year. Its just reinforces the fact that 1999 was obviously an excellent year for long-term systems.
No not obvious-- I think its more a case that in 98 you had open profit which carried through and exited in 99.
Run your equity curves with open profit shown,gives you a better idea.

Often with longer systems the early years will "look" terrible when in fact they are brilliant with massive un realised profit hiddedn in the open trades.
This is more often the case for poor "appearing" results than actual poor performance!
 
Joined
Dec 15, 2006
Posts
8
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

MS + Tradesim,

Is the filter you use Jose's? Guessing also that trades exit based on an index also?

Thanks.

(PS Been a lurker for a few years, first post I think?)
 
Joined
Jun 14, 2007
Posts
1,130
Reactions
3
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi Plan-trader,

Yes, it's an application of Jose's URSC kit. Exits are based on a trailing stop only - unrelated to the index and unrelated to all the usual volatility based trails. I spent over a year refining it, so it stays secret :) I also have his MACDH kit but have not yet really spent time with it. Hoping it will be useful with FX.


Hi Chorlton,

As tech says, long-term systems can have poor initial years simply because the non-performing trades are being closed out while the good ones just keep going. I do agree though that a good starting year will help.

MaxDD is around 17%. Starting capital is 100k, however it's unconditionally profitable starting with as low as 15k - variance just continues to rise the lower the starting cap. Profits are compounded, risk is 2%.

It's just a coincidence that the filter doesn't work prior to 2000. Seems to be to do with how Jose set it up. However, every system I've created running from 00-09 is improved dramatically by applying it, so I have no doubt that it would also have helped prior to 2000 were it available.

Working on SPI and FX systems at the moment so still haven't got back to testing B/E stops. :eek:
 
Joined
Apr 14, 2007
Posts
317
Reactions
0
Re: Difference in Win % : Short-Term vs Longer-Term Systematic Trading

Hi Plan-trader,

Yes, it's an application of Jose's URSC kit. Exits are based on a trailing stop only - unrelated to the index and unrelated to all the usual volatility based trails. I spent over a year refining it, so it stays secret :) I also have his MACDH kit but have not yet really spent time with it. Hoping it will be useful with FX.


Hi Chorlton,

As tech says, long-term systems can have poor initial years simply because the non-performing trades are being closed out while the good ones just keep going. I do agree though that a good starting year will help.

MaxDD is around 17%. Starting capital is 100k, however it's unconditionally profitable starting with as low as 15k - variance just continues to rise the lower the starting cap. Profits are compounded, risk is 2%.

It's just a coincidence that the filter doesn't work prior to 2000. Seems to be to do with how Jose set it up. However, every system I've created running from 00-09 is improved dramatically by applying it, so I have no doubt that it would also have helped prior to 2000 were it available.

Working on SPI and FX systems at the moment so still haven't got back to testing B/E stops. :eek:
Hi MA+TradeSim,

Thanks for taking the time to reply. I too, have spent a long time playing around with exit conditions. When I look back at my AB code, the exit conditions make up around 75% of the entire system code !! ... and I'm still adding to them... :(

All the best,

Chorlton
 
Top