Monday, August 18, 2008

The Art of Risk Control--Position Sizing. Part I

The topic I'm going to share today is a departure from value investing -- position sizing. Why position sizing you may ask? This is a very powerful risk control tool to prevent damage to your capital. This is one of the major difference between a pro and an amateur.

Assuming I have RM 10,000 capital. If I bought 2000 shares of IOI Corp at RM 4.70 cost me RM 9,400. I am committing 94% of my capital into one stock. If IOI Corp were to drop by 25%, I'm losing my capital by RM 2,350(0.25*9400). In this case I have lost 24% of my capital, left with RM 7,650 after taken the losses. I need about 31% return to get back my capital on my next investment. If I'm lucky, at 10% return per year, this will take me about 3 years to recover my principal!!!!!

Position sizing is something that many value investors do not concern too much because they think they are protected by margin of safety ( low PE, low book value, etc), they therefore think committing 94% of their capital is something that they feel comfortable and safe. They may even think the possibility of their stock to go down by 30% is unlikely. Well, you may think it is unlikely when CPO is at RM 2500/MT, how about CPO drop to RM 1,500?

There are several strategies to deal with position sizing. However, my favorite is percentage risk model. Basically, you must determine how much capital you are willing to put at risk and your stop loss percentage.

In the following example, I am willing to lose 3% of my capital with a stop loss of 25% for IOI Corp. You may construct this simple calculator in a Spreadsheet.



To lose no more than 3% of my capital with a stop loss of 25%, I am allowed to commit only RM 1,200 to buy IOI Corp or 12% of my capital. If I realize my losses, this is no big deal as I have lost only RM 300. I still have RM 9,700 of my principal, with 10% per annum return, it will only take me only about 3 months to recover my principal.

Can you see the difference?

Now you may ask me why am I committing such a big position on my MUI and Parkson? When I started with such small position, it is inevitable that I'm exposing myself to take a lot more risks. When I took at that size, I'm taking into consideration of my future cash flow, with RM $ 888/month, I know I am going to have about RM 10,000 in the first year, I'm using this forward capital in my sizing consideration. With RM 2,500 per position or investment idea, I am exposing myself a potential 6% losses on my capital on 25% stop-loss. I think I can live with that based on the size of my stomach since this is still less than 10% -- but I think my risk management should be OK.

If you want to learn more about position sizing, Trader Mike wrote an excellence piece on this topic, you may visit http://tradermike.net/2005/07/position_sizing/

Good luck.

2 comments:

Halim said...

Excellent post!

After a lot of backtesting of KLSE, I have come to this conclusion: "Assuming you have a positive expectancy (i.e. a trading system that works) it is money management (which includes position sizing) that's going to make the difference between ruin or mammoth returns".

Even with random entries, it is position sizing and exits thats going to make the cut.

I personally risk 2% capital with 2*ATR initial stop loss.

James said...

Halim,

I agree. Perhaps you can help me refine my sizing algorythm based on your [and my] choice of 2% of capital risk with 2*ATR stop. I've been trying to backtest a system with thse conditions using AmiBroker. The results started driving me crazy when it would buy huge positions that I would never consciously buy myself. Thinking my coding was buggy, I dug deeper. My code was OK[I think], but here is an example of the problem. Assume total capital is 100000. On 5/20/2008 SHY closed at 83.22 and its ATR(15) was 0.20. Please correct me if I'm wrong but I calculate PositionSize as 100000*0.02*83.22/(2*0.20) which is 416100, which is 4 times available capital! When I program other limits the system loses continuity and becomes very difficult to optimize. I want PositionSize to increase on issues like SHY which have low relative volatility, but not by this much! Any ideas?

Thanks,
--James