Position Sizing with Concurrent Signals and Portfolio Risk
Summary
The discussion addresses sizing long and short stock positions generated by a daily strategy when trades overlap and signals arrive unevenly. One proposed approach budgets exposure using the number of concurrent long and short positions. If a signal has an associated probability or confidence estimate, that information can also inform the size, through a meta-labeling framework.
A second approach treats sizing as part of portfolio construction and risk management. It recommends examining historical portfolio volatility and drawdown, then relating those outcomes to the strategy’s scaling coefficient, the volatility of each stock, and correlations among positions. Estimated volatilities and covariance can guide adjustments to the coefficient. The thread also points readers toward asset allocation and portfolio construction research. It offers conceptual starting points and references, but no comparative tests or universally optimal sizing rule; suitability depends on the strategy’s signals and estimated risks.
Key ideas
- Concurrent long and short positions can inform a budget for allocating capital.
- Signal confidence or probability can be used to vary individual position sizes.
- Historical portfolio volatility and drawdown help assess the risk of a sizing rule.
- Position scaling should account for asset volatility and correlations between holdings.
- Asset allocation and portfolio construction research provide further starting points.
Tags
Full text
# Position sizing in algorithmic trading # Position sizing in algorithmic trading Good morning, I have a question, regarding position size in algorithmic trading. I have a strategy that every day generates signals for buying or selling positions on different stocks. I'm looking for the hints and advices in order to manage and optimize position sizing of my strategy, where can I start my research from? I tried with portfolio management and risk management but they usually refer to a different scenario. Note that I hold these position for about 20 days and that signals are not evenly distributed in time. For example, you can have 20 signals in a day and zero signals for a week. Thanks ## Answer by kingfischer (score 5) https://quant.stackexchange.com/a/46924 If you already have a strategy generating potential long and short positions, you may want to check Chapter 10 of Marcos López de Prado's book Advances in Financial Machine Learning. It describes a number of strategies that include a budgeting approach where you only rely on the number of concurrent long and short positions to optimize position sizing. If you have the probability or confidence in a particular position you can use that to derive a position size as well (he calls this the meta-labeling approach). I've created a Jupyter Notebook that implements his sizing strategies and examples here. It is part of a larger project by a number of people to implement and continue de Prado's work, and bet sizing will be released as part of an update to the mlfinlab Python package ## Answer by lehalle (score 2) https://quant.stackexchange.com/a/46895 You have to consider seriously to use portfolio construction, at least for risk management purposes. To start with, you should "backtest" the risk you are taking with your current "proportional strategy". What would have been: - the volatility of your portfolio (ie your aggregated positions) in the past? - your maximum loss (max draw down) in the past with this strategy? - etc. you will see that it is a function of: - the proportionality coefficient you have chosen - the volatility of the stocks your are taking positions on - the correlations between your positions Portfolio construction is simply about adjusting the coefficient to the two other quantities, based on reliable estimated of the volatilities and the covariance. You should really do it. ## Answer by amdopt (score 1) https://quant.stackexchange.com/a/46920 > where can I start my research from? Searching for asset allocation papers and portfolio construction papers will yield far better results. Here are a few papers you may find interesting to get you started: Optimal Trade Sizing in a Game with Favourable Odds: The Stock Market A Quantitative Approach to Tactical Asset Allocation
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.