different % Risk for longs and shorts

Discussions about Money Management and Risk Control.
Post Reply
cordura21
Roundtable Fellow
Roundtable Fellow
Posts: 61
Joined: Thu Mar 11, 2010 12:45 pm

different % Risk for longs and shorts

Post by cordura21 » Wed Sep 22, 2010 11:57 am

Hey guys. What's your opinion on using a different % risk on equity with long versus short trades? Any practicioner?
Let's assume that shorts are faster than longs (a test on emerging markets showed me that for the decade short movements are 2/3 bigs than longs, and they last half of the time). If you use the ATR to make for a tighter stop, then positions get too big in relation with your equity. So another alternative is to adjust the risk.
I am asking this because I don't see it often on systems or literature, but I could be missing something very important. And I am also aware that, by adding more parameters, it could be seen as curve fitting (although I saw some systems that leverage themselves on the upside and not on the downside).
Please share your ideas about this.
Regards, Cord

Moto moto
Roundtable Knight
Roundtable Knight
Posts: 427
Joined: Mon Jun 01, 2009 4:12 am
Location: once again in the UK

Post by Moto moto » Wed Sep 22, 2010 12:55 pm

possibly make sense in equities, but does it make sense in bonds, currencies or commodities?
test it.....:)
regards FX I have seen a few traders say they only like to short currencies, to me this is a nonsense unless you happen to base everything in your home currency eg; USD and then only trade other currencies against it. what about then GBPAUD , EURJPY etc.... or is it AUDGBP, JPYEUR.
Another alternative depending on the overall rationale for having a separate idea for longs v shorts is rather than adjusting the percentage risked, adjust where you exit a position ???
in terms of curve fitting - is it curve fitting if you have an alternative theory to the markets and then you apply a separate method to trade the longs v the shorts... I dont think so, so long as you are not optimising and curve fitting for parameters.

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Wed Sep 22, 2010 1:09 pm

There have been a number of discussions on this forum about the merits (or otherwise) of shorting. I fear you will have to search them out but certainly Forum member Sluggo has made a number of suggestions in this area which include (from memory): 1) not shorting at all; 2) adjusting per trade short risk to some lesser figure than per trade long risk.

Some traders reach the decision that shorting, while minimally profitable for most/many TF strategies, is worth it in terms of smoothing the equity curve. Some reach the decision that taking on board the risk is not worth it, especially if they operate (as I imagine most do) overall and sector risk control. Most of the time short trades take up an allocation of risk in return for very little profit.

It is however worth analyzing different time periods. Without being too precise about it, I recall looking at shorting strategies towards the beginning of the period where I have data for a meaningful number of futures contracts - the 1970s through to the early 1980s. During long term declines in prices of certain sectors, a long only policy can reap scant or negative rewards if there are not enough sectors in uptrend and repeated attempts at long trades bring only attrition of capital.

More recent years have brought much wider diversity in trading sectors, meaning that it is more likely that you will find long only opportunities - more likely that you will find some group of instruments going up. The early days were mostly grain contracts.

Nowadays, if grains are going down as a group, perhaps some other sector will be going up. If stocks are going down, perhaps bonds, STIRs and gold are going up.

Markets do change, strategies need to change with them. Shorting has not been a great game over the past couple of decades (on balance). But a prolonged period of deflation could change the picture.

As ever, there are no definitive answers, no way to predict the future. The only secure knowledge is that the future is going to be different in many respects, even if it will share some similarities with the past.

In practical terms this means persevering with our trading and keeping a very open mind as we continue to observe, back test, research and ponder. No one has the answer but those who constantly observe and question and move with the times have a chance of adapting to changing conditions.

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Wed Sep 22, 2010 2:57 pm

Some of the possibilities (these are not necessarily recommendations) might include
  1. Don't trade short at all
  2. Trade short but at a smaller per-trade risk than long trades
  3. Trade short at the same per-trade risk as long, but set Max_#_Short_Positions smaller than Max_#_Long_Positions
  4. Trade short at the same per-trade risk as long, but set Max_Total_Heat_Short smaller than Max_Total_Heat_Long
  5. Trade short at the same per-trade risk as long, but scale out from shorts sooner than from longs (tighter profit targets)
  6. Trade short at the same per-unit risk as long, but don't pyramid shorts as many times (smaller max#units)
  7. Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors, on the theory that this is pure anti-correlation to the S&P benchmark, and why not grab big handfuls of it.

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Wed Sep 22, 2010 3:09 pm

Those who own Trading Blox testing software can easily test the idea of trading Long-Only, or Short-Only, or Long-and-Short. The software provides a handy "Portfolio Manager Block" called the Trade Direction Portfolio Manager. Drop it into your system (which takes all of fifteen seconds) and away you go.

I just did this very thing to one of the trading systems I enjoy testing, and ran the Blox program three times. I captured the Yearly Performance Summary (by right-clicking on the table and choosing "Export to Excel") of the three runs, and then combined them together. Here's what I found for this particular system and this particular portfolio.

As you look at the table, ask yourself "Which looks best to me? Long+Short? LongOnly? ShortOnly?"

The other attachment is a plot of the three equity curves. It may surprise you. You might choose one option when viewing the yearly performance table, and then change your mind after you see the equity curve plot. (That's why I intentionally put the equity curve in a .zip archive, so your eyes wouldn't be polluted by the equity curves. It gives you the opportunity to study the table and choose an option, before seeing the equity curves.)
Attachments
Ecurve_Plot.zip
Equity curves of trading this system Long+Short, LongOnly, ShortOnly
(22.02 KiB) Downloaded 339 times
Trade_Direction_Portfolio_Mgr.png
Yearly Performance Summary, trading this system Long+Short, LongOnly, ShortOnly
Trade_Direction_Portfolio_Mgr.png (28.16 KiB) Viewed 6430 times

LeviF
Roundtable Knight
Roundtable Knight
Posts: 1428
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF » Wed Sep 22, 2010 3:30 pm


AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Wed Sep 22, 2010 4:28 pm

Happily, it is not a moral decision: there is no "right" or "wrong". Justs systems to be tested and decisions to be made. The test below shows that for the system, portfolio and test period, long only would have produced greater profit. Who knows what the future may bring?
Attachments
Chart.JPG
Chart.JPG (71.08 KiB) Viewed 6399 times
Table.JPG
Table.JPG (78.35 KiB) Viewed 6398 times

cordura21
Roundtable Fellow
Roundtable Fellow
Posts: 61
Joined: Thu Mar 11, 2010 12:45 pm

Post by cordura21 » Fri Sep 24, 2010 2:02 pm

Thanks everybody for the thoughts. Since then I've been tryng stuff. On a breakout system with stocks, shorting stocks takes the MAR from 0.6 to 1, and it makes both the CAGR and the maxDD bigger.
I also tried using half the risk per trade and the stop distance with shorts, and the MAR improved another 0.20 points.
Is my rationale logic: since I halving my stop my positions should be bigger, but at the same time my risk per trade gets smaller, so the effect is that now I am less tolerant to pullbacks on the short side.

Sluggo: on point 7 of your comments "Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors, on the theory that this is pure anti-correlation to the S&P benchmark, and why not grab big handfuls of it."

You mean "sectors" as in different asset classes or sectors as subclasses of the benchmark? Can you explain it a little more? Thanks a lot

nodoodahs
Roundtable Knight
Roundtable Knight
Posts: 218
Joined: Wed Aug 09, 2006 4:01 pm

Post by nodoodahs » Fri Sep 24, 2010 5:56 pm

I would guess from context that he was using "sectors" in the same way I would have used the term "asset classes" to describe futures contracts on different types of instrument.

- stock index futures
- physical commodity futures (could be further subdivided)
- bond futures
- currency futures

Grabbing anti-correlation is a reference to using a futures program as an addition to a larger portfolio e.g. a pension fund might do this, and grabbing some outperformance by shorting stock index futures with gusto might help the overall performance of the portfolio. It doesn't necessarily do anything good or bad (you'd have to test it LOL) when you're using the futures program as your only portfolio.

Back to the idea of sizing bets long and short, another avenue to explore is Kelly Betting to set the sizes of each, then combine them. A simpler approach would be backtesting various sizes on each side (long and short), and seeing if they didn't optimize separately to different sizes.

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Fri Sep 24, 2010 7:07 pm

Original: 7. Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors, on the theory that this is pure anti-correlation to the S&P benchmark, and why not grab big handfuls of it.


Spelling out the details more explicitly: 7. Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors (which are: grain futures, metal futures, livestock futures, energy futures, short term interest rate futures, long term interest rate futures, currency futures, and tropicals/softs futures), on the theory that shorting stock index futures provides pure anti-correlation to the buy-and-hold-the-S&P-500 benchmark, and why not grab big handfuls of anti-correlation to the benchmark.

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Sat Sep 25, 2010 2:34 am

Straying from the topic somewhat, I recently put together the following dumb little bar chart by taking every down month in the S&P 500 over the past decade, adding it and comparing it to the same cumulative total of monthly performance in 5 single manager funds (including our own) and also in a little fund of TF funds we are putting together.

Manager 1 trades stock indices long only. Managers 2 and 3 do not trade stock indices at all. Managers 4 and 5 trade stock indices both long and short.

Apropos of nothing in particular it is interesting to note that managers 2 and 3 seem to provide the most support in equity down months which is not uber evident from the simple monthly correlations over the period.
Attachments
Corel.JPG
Corel.JPG (20.01 KiB) Viewed 6258 times
UpDown.JPG
UpDown.JPG (32.82 KiB) Viewed 6260 times

cordura21
Roundtable Fellow
Roundtable Fellow
Posts: 61
Joined: Thu Mar 11, 2010 12:45 pm

Post by cordura21 » Mon Sep 27, 2010 11:35 am

Thanks for the clarification, Sluggo. I guess it could be a matter of getting anticorrelation by trading/timing versus by asset allocation, which is supposed to bring anticorrelation by itself.

I think that depending on the "client" (be it yourself or another person), comparing to the benchmark can have an effect that could affect allocation. If you compare too much to the index, you'll be comfortable on periods when the benchmark is down, and kind of anxious when the benchmark is above your curve (some people say that the second is much more uncomfortable than the first situation).

In cases where you're winning, and you are aware of these tendencies, you could be tempted to switch to follow the index more closely, just to avoid being in the situation where you missed a move. Now I know this is far from an ideal behaviour and that as systematic traders, is something you are decided to avoid. But don't you think this is sometimes the case?

sluggo wrote:Original: 7. Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors, on the theory that this is pure anti-correlation to the S&P benchmark, and why not grab big handfuls of it.


Spelling out the details more explicitly: 7. Treat different sectors differently; for example, allow bigger position size when shorting stock index futures than when shorting the other sectors (which are: grain futures, metal futures, livestock futures, energy futures, short term interest rate futures, long term interest rate futures, currency futures, and tropicals/softs futures), on the theory that shorting stock index futures provides pure anti-correlation to the buy-and-hold-the-S&P-500 benchmark, and why not grab big handfuls of anti-correlation to the benchmark.

Post Reply