The Trade-Robots Strategy Portfolio - 2014-09
3 People. 4 Years. 12 Man years. That is a lot of time, especially when all you do is strategy development. Spend 90% on strategy development. We do our own infrastructure (but that is separate topic) and optimized everything for speed of development. If you put in 4 man years and a software package that enables us to be maybe 1000 times as efficient as NinjaTrader on our combined hardware – something is bound to happen.
In our case, it is out “Wall of Strategies”. An overview of every single strategy that we are following in order to consider it for the allocation for funds, or that is being traded in cash. We are funded – not too bad – but this is in some sort a wall of shame. It shows a lot of things we could do with more funding for our trading accounts.
Right now, most of these strategies live there on the wall. Once per month they get an update when we update their backtests with the newest data, a manual process we plan to increase to a weekly schedule. This is not optimization – we just collect more data and see how they work in the future. Right now we cannot do automatic walk forward optimization. So we just track the “future” performance of those strategies. Any of those strategies has been decided to be worthy. Some trade cash. None is allocated enough funds to use its potential.
Will we make the available for trading by investors? Yes and no, possibly. Yes, but not in this form. This wall is purely for us to see how much we still have to do. We think about turning them into portfolios. Combining them, making one of more portfolios that are optimized to reduce volatility. Single signal / strategy optimization is more limited than taking for example 3 markets, 3 time frames and 3 strategy types – one trend following, one counter trend, one mean reversal. The result, when done correctly, is reducing volatility.
Strategies we develop at NetTecture are quite sophisticated code wise, but we follow a very strict “keep them simple trading wise” approach. Yes, many lines of code for trading etc., but we normally isolate separate signals into separate strategies. This is contrary to what we know of other traders. They run very complex strategies that have multiple trading signals. This is a distinction we plan doing on the portfolio trading level. Right now manually, definitely for anything we offer to any others, and maybe in an automated MultiStrategyTrader module for our Reflexo framework in 2015. For now, we manually combine strategy instances we analyse to be smoothing the equity curve.
What is your situation? Do you have a lot of strategies that just lack the funds to trade them? Let us know.