tradeQuantiles: quantiles of per-trade stats

In braverock/blotter: Tools for Transaction-Oriented Trading Systems P&L

quantiles of per-trade stats

Description

The quantiles of your trade statistics get to the heart of quantitatively setting rational stops and possibly even profit taking targets for a trading strategy or system. When applied to theoretical trades from a backtest, they may help to adjust parameters prior to trying the strategy with real money. When applied to real historical trades, they should help in examining what is working and where there is room for improvement in a trading system or strategy.

Usage

tradeQuantiles(
  Portfolio,
  Symbol,
  ...,
  scale = c("cash", "percent", "tick"),
  probs = c(0.5, 0.75, 0.9, 0.95, 0.99, 1)
)

Arguments

Portfolio	string identifying the portfolio
Symbol	string identifying the symbol to examin trades for. If missing, the first symbol found in the Portfolio portfolio will be used
...	any other passthrough parameters
scale	string specifying 'cash', or 'percent' for percentage of investment, or 'tick'
probs	vector of probabilities for quantile

Details

This function will use the quantile function to calculate quantiles of per-trade net P&L, MAE, and MFE using the output from perTradeStats. These quantiles are chosen by the probs parameter and will be calculated for one or all of 'cash','percent',or 'tick', controlled by the scale argument. Quantiles will be calculated separately for trades that end positive (gains) and trades that end negative (losses), and will be denoted 'pos' and 'neg',respectively.

Additionally, this function will return the MAE with respect to the maximum cumulative P&L achieved for each scale you request. Tomasini&Jaekle recommend plotting MAE or MFE with respect to cumulative P&L and choosing a stop or profit target in the 'stable region'. The reported max should help the user to locate the stable region, perhaps mechanically. There is room for improvement here, but this should give the user information to work with in addition to the raw quantiles. For example, it may make more sense to use the max of a loess or kernel or other non-linear fit as the target point.

Author(s)

Brian G. Peterson

References

Tomasini, E. and Jaekle, U. Trading Systems - A new approach to system development and portfolio optimisation (ISBN 978-1-905641-79-6)

See Also

tradeStats

braverock/blotter documentation built on Dec. 16, 2024, 1:02 p.m.