I said if there is an edge TSL and APS may be able to find it because these programs can discover completely unknown rules that are not combination of other known rules. One is still required to recognize that is an edge and this takes a lot of experience and effort. I also said that I do not believe SS can provide an edge because what it basically does is rule permutation. I believe that in the past 20 years every conceivable trading rule based on known indicators and technical analysis that could pinpoint to a market anomaly has been exploited by traders and then faded by the markets. If you can come up with an edge and you need to backtest it then any program can do the job, like SS, Amibroker, Metastock, Tradestation, etc. There is nothing special about SS in that regard. The true and only reason - besides poor money and risk management - that 95% of all traders lose money is that they fail to come up with an edge. They backtest a system, it works nicely for a few months and then a huge drawdown comes along.
IBill, You are being somewhat naive if you do not think "order flow" is massisve edge,and "somewhat" is being kind.. Keep in mind I was referring to investment banks.. As for SS vs the others you mentioned,it is vastly different,and I do own the other programs.
Rule permutation systems can be profitable over several years, but there are definitely obstacles to overcome to trade them successfully. My knowledge about APS is limited to what they published in a paper in late 2006. The front end of their basic process appears similar to that used by Stratasearch in that seed rules are used as search input.
On the contrary, I think anyone who believes order flow can provide a massive edge understands nothing about order flow. I get the impression you are an old timer who worked during the time order flow in a pit occasionally provided some edge to locals and their firms. However, retail traders like us were never able to take advantage of that. Now a days, with electronic direct access and orders like "iceberg", ability to instantly remove an order and bluff, order flow cannot provide an edge. Actually, that was part of the initiative to make markets more efficient by removing any edge due to order flow knowledge. If you do not understand that markets have gotten more efficient from that perspective I have nothing more to say.
I exchanged a few emails wwith the staff at tradingpatterns.com about this and I was told that this is not the case and there are no seed rules in APS. TSL uses seed rules but employs a genetic programming algorithm that can create completely new rules. Having used APS for a while I get the impression that it monitors price action of up to 6 days and configures the rules based on that. I don't know how sophisticated their algorithm is but it is not based on seed rules but on something they call "general pattern groups", whch they differentiate from rules: http://www.tradingpatterns.com/Support/UseAPS/useaps.html#usefaq21 "There are no hard coded patterns in APS. Instead, APS uses algorithmic "data mining" guided by a set of up to 88 (extended search) predefined general pattern groups. These 88 groups are not exact patterns but correspond to thousands of exact pattern formations. " It's kind of confusing to me and it sounds like they are not willing to reveal the way this works in more detail. This is understandable.
Bill, I was the head trader of equity derivatives/emerging market equity derivatives at a major IB..which is still in business I now understand where you are coming from.. But you are way off the market in your assessment of "flow"...
I don't know whether an edge is possible via these programs but I tried the Stratasearch demo with the pattern add-on. After 3 days the program was still running and I had to temrinate it because it seemed stuck in an endless loop. Using the same data file and parameters it took APS less that 11 minutes to execute a search and come up with results, about 27 patterns in total. You can do some test yourself as both companies offer a free demo. This is not to say that I support the view that any program can provide an edge. My comments have to do with execution speed only. I believe the true edge is in risk management and ability to understand when to trade and when not to, some kind of higher level money management I guess. John
I was able to get the same APS results using the data files one can download from this webpage and the APS demo version: http://www.tradingpatterns.com/Software/APS/forex/forex.html Each search took a few minutes, I would say less that 10. On the other hand, SS with the pattern search library and with only one of the files was running continuously for a few days before I had to shut it down because I had to make a hardware change to my PC and never saw any results. Probably SS was in an infinite loop. I can't explain it otherwise. SS people on their website forum attack APS as a simple program but it seems their own program cannot do what APS does quite fast.
APS looks simple because of the simple interface but it's not a simple program. SS is kind of confusing to use. It is certainly a good product but I agree it just takes too long to run. Some statistics based on my testing: Search domain: OHLC patterns APS extended pattern search run: 1 hour 12 min SS OHLC pattern search run: 3 days + (terminated) Results: APS appears to filter out random patterns. SS results appear to be mostly random selections of O,H,L,C points. Conclusion: APS search is based on some underline logic or strategy. SS search is based on examining all possible permutations and thus subject to randomness. Ron