The leverage space method requires knowing future drawdown levels. It is impractical as recent market history tells. Never risk more than 2%...
Yes, the Leverage Space Method makes the assumption that the future behavior of your system will be similar to the past, i.e. it use the past returns as an estimate for the future (much in the same way as volatility estimators). So NEVER use backtest returns as input; either use walk-forward data or real returns. Also, nobody forces you to leverage your system as indicated by LSP. The method also assumes that you can scale out as you get into drawdown. This means that your âreaction timeâ also impacts the end result. What is not really described in the book is how to handle the estimation risk (mentioned above) that you take by using historical returns of the system. You can argue that the estimate is more precise the longer the series is. But LSP only estimates the risk of hitting your drawdown limit provided that historical returns is a good estimator of future returns. If the system really isn't as good as your historical data suggests LSP will suggest too high leverage. So what happens if the system stops working? Ideally, as you reach your drawdown limit, the leverage for the system will approach 0. In practice you need to take other things into account, e.g. overnight risk, tail risks etc. Even so I believe the LSP method provides a good balance between theory and practice, i.e. how to maximize returns given a certain drawdown risk. In the end all comes down to how you think about risk. My view is that you should try exploit a functioning trading system as much as possible given the risk you are prepared to take.
I don't know the details about LTCM but wasn't their problem that they added to the positions (i.e. increased leverage) in drawdown believing more in their theory than the current P/L? LSP can be set up in a lot of different ways, but as far as I know it will scale out when in drawdown (if close enough to the drawdown limit). Of course you can have concerns related to the theory behind LSP (optimal f and drawdown risk) but IMO it is better to understand the risk using a theory than just putting your finger up in the air.
Look at the GLD chart for the last two days.For a trend follower or position trader who used LSP for position sizing chances are he just got killed. It appear that the optimal position sizing method is to risk as little as possible.
I understand your position but personally I'm more in the camp advocated by Aaron Brown (writer and contributor at NuclearPhynance) - risk management is not about avoiding risk; it is about understanding the risk you take, define the risk you are willing to take and to manage the risk to get the best returns you can from your systems. This is no easy task and there is an obvious model risk as discussed earlier.
Aaron Brown is a very smart man I agree. He knows a lot. But, think of other possibilities. Try to risk very little more frequently. There is no obvious answer as you claim. It is a matter what fits your objectives best. What do you prefer? A - Risking 2% - 5 times a month? or B - Risking an optimal 30% (for example) once a month? I understand A is difficult to do in terms of entries but I wouldn't go for B. Better not to trade at all. Someone is waiting to take your 30% be sure about that. It is called the "big fish".
With the same expected return in both cases then of course A. And yes, Aaron seems to always have good ideas. As a side note, we normally use the risk level "less than a 10% chance of reaching a 20% drawdown in the next 12 months". This results in modest leverage for most of our systems. How to handle risk also depends on the type of system. For example, a system that is continously invested in a number of stocks seems to be easier to manage with LSP type of methods than systems that looks for very specific set-ups in futures (it seems like quite a lot of traders run these systems, looking for special situations and then leverage it to the hilt in order to get decent monthly returns, often leading to disaster). IMO systems with infrequent, often high, exposures like this are harder to analyze correctly. The requirements of reaction time, the validity of previous returns etc all seems to make it tougher. In the end the most important thing is to try to understand the risks you take and the problems your sizing method may have.