I tried an advanced search with the following:
Instruments traded: Futures, Stocks, Options, Forex
Speed: Don’t care
Trades: 30 min num trades
W:L > 1.5
Sharpe > 0.7 (good)
$win/share 50 min $ profit per contract/share
I came out with only 2 systems: Midas Long-Term Value and Midas Short-Term Value.
My question is: Is this correct? There are only 2 out of over 500 systems at C2 that meets this basic minimum criteria or is there something wrong with the search algorithm? It is important that we know this because if there is something wrong with the search algorithm, we need to certainly fix it.
rgds, Pal
Midas Long-Term Value
Midas Short-Term Value
Also, I’m unable to save a search with anything in the $win / share field. It simply refuses to save a search with anything in that field and returns the search with nothing in that field and the results accordingly when cliking on the search name.
rgds, Pal
Pal - This is now fixed. Thanks for pointing out the problem. - MK
I think the search algorithm is working okay. The problem isn’t with the algorithm; it’s with the trading systems. Most aren’t very good. - MK
I suspect also that the $win/share is not really a valid parameter for assessing the value of a system. Highly leveraged markets (like forex or futures) will generally have higher per share values than stocks - but most investors aren’t concerned with how many shares they buy/sell - they are concerned with how much capital is at risk for a given reward. That is why W:L (or Profit Factor to us techies) is a better metric - it tells you how many dollars you gain for every dollar you lose - much clearer to a money manager/investor than $ per share. Think about it: If you are trading .50 stocks and looking for a .05 profit then your $/Share is 5 cents! But your profit is 10%! Conversely, if the shares are $200 and you target a $5 gain, then your $/Share is 5, but your percentage gain is only 2.5%. $/Share is 100x better on the latter, but profitability is only 25% of the former. You might consider something like Average Trade; It tells the investor the average value of a transaction on the account (given 100k capital) and is independent of the underlying instrument and risk profile.
You’re right that Average Trade would be a good statistic to show (and so I’ll add it). My opinion about “dollars per share/contract” is that it is not a useful statistic for comparing trading systems across multiple instruments (i.e. futures systems, stocks systems), but is pretty useful when comparing within an instrument class – and particularly when looking only at stock-trading systems.
My thinking is this: if the average dollar win per stock share is something like $0.05, then you know that the system won’t be very good in real life – what with commissions (.02 per share per round trip) and slippage (varies, but clearly it will be a big percent of the remaining .03 cents).
Ultimately, I intend to add a “Realism Factor” – which will describe how likely a trader is to receive the same profits as appears on Collective2. It will factor in things like trading volume and liquidity, trade frequency, commissions, etc. But that’s a bit further down the road.
Matthew
Well, stock options are highly leveraged instruments too. Nothing prevents these systems that trade only stocks from not trading stock options also, either to hedge their portfolio or to boost their $win per share/contract. Afterall, as somebody stated, (stock, futures, forex) options are governed primarily by the movement of the underlying (stock, futures contract, forex pair).
Ultimately it comes down to the payoff structure (robustness) of the portfolio and its sensitivities to various market factors and economic shocks and the closest statistic we have right now at C2 to quantify “robustness” is this $win per share/contract statistic and so it is in my opinion a key statistic and a real eye opener.
rgds, Pal
Ok. Thanks for your prompt reply and prompt bug fixing. Most trading systems don’t work well when tested in hypothetical real-time (walk-forward) testing, but most of them do work well with a good money management system.
I suspect that most trading systems at C2 don’t have a good money management system, with the result that we have such poor results as clearly evident in the $win per contract/share statistic, particularly more illuminating, if combined with a good W:L Ratio (Profit factor), min. # of trades (30 to be statistically valid), and a good Sharpe Ratio.
Thanks for a great advanced search tool. It sure exceeded my expectations inspite of some small functional differeences with what I expected intuitively. As usual, MK, you amaze me more and more as time progresses. As somebody was saying, “genius” is the capacity to take pains.
rgds, Pal
