Machine designed strategies. Do they work?
- Thread starter v75z52
- Start date
Yes, there is a lot of leverage available out there, but the rationale for selling is the timing aspect of it. Again, for someone who is capital-constrained, the ability to instantaneously monetize the intellectual capital embedded in the software can make sense.
So, yes, I would never, ever, ever buy software from someone as a retail trader, but as a strategy developer I MIGHT be willing to give up some percentage of the future gains (as well as offload some of the risk of the strategy ceasing to work) in exchange for access to the strategy.
Let's say some hedge fund heard of your strategy and said they would give you $1 million today for access to it because they think they can make more than that implementing it with their capital. You know, as much as these things can be known, that you could, in 10 years, make $1 million in net present value terms, but you'd rather have the money now than wait (time preferences for liquidity are fairly idiosyncratic, after all). That strikes me as a rationale for selling profitable software.
Of course, my original point was that since this is not the kind of deal that software vendors catering to the retail market are offering, that they are basically rip-offs. If they really believed their strategies worked, they might sell access to it, but the price would be some type of gain-sharing, not an outright purchase. I actually turned down an offer by someone with the resources to automate my strategy precisely because he then wanted to sell it, whereas I wanted to continue trading it without giving up some of the market to the buyers, on the rationale that there is only so much liquidity to go around and I didn't want others to be sucking it up trading the same algorithm.
So, my bottom line would be don't buy software unless the seller asks for a piece of the ongoing profits because that's a sign that the seller really thinks the strategy works, they just want to scale it up more than they can do on their own.
On the other hand, if a buyer believes the system works, it's better for him to pay a one off price, than share the profits.
It's a little off-topic, but what is even the definition of "efficient" or "inefficient"? My opinion is that the market is always efficient and always inefficient. It's efficient because at that moment, price is reflective of all available information ("it is what it is") and it's inefficient because price is never static. If there is one thing I've learned from watching the market over the years, it's that the market is its own universe (I cannot overemphasize this enough, actually. If I could have known one thing when I started watching the markets, I would choose to have known this.) and expecting it to be understandable on anything other than its own terms is futile. Just as light is both a wave and a particle, the market is both efficient and inefficient.
As for whether something very complex can be encapsulated in something fairly simple, I would refer you to Einstein's E=MC2 or Netwon's F=MA, both of which are simple equations which explain vast amounts of the known world. Having written code back in the early 2000's, I would estimate that my strategy, fully-coded, would be a few hundred, may up to 1000, lines long. Long enough that it isn't overly simplistic, but not so long that it basically ends up overly complicated and brittle.
Right, but it makes sense that the seller in this case has the greater power in the negotiation and can set the terms, since the seller is the one with the asset the buyer wants.
If a hedge fund offered me a few million bucks, I would explain my entire strategy to them and then we would agree to share the market in the assets where it worked. Since the strategy has an objective answer of when to trade, how to manage the trade, etc., I would easily be able to determine if someone else was using the exact same strategy. I actually often wonder if anyone is, since clearly there are other transactions happening at the same price as mine and to some extent I think of trading strategies as being akin to the "million monkeys on a million typewriters", one of which eventually types out Shakespeare's complete works.
I know my strategy is profitable, what I don't know is if anyone else knows it, too, or why it happened that I was the person to discover the strategy. I guess to some extent it's because only I had the experiences I've had and without that exact sequence of experiences, the strategy would not be discoverable. I keep saying to myself that it will eventually stop working, but it just keeps on going. At worst, the strategy stays out of the market and the frequency of trades declines. In terms of actual drawdown, though, it never really has a major one.
That's a whole another topic. I always assume that someone somewhere uses the same strategy as mine. Millions of people search for profitable strategies and it's very unlikely that I'm the only one who discovered my strategy. So to some extent it doesn't matter if my strategy is known publicly or not, I always assume that it is. After all, there are thousands of trading sites on the Internet that I've never visited and they could be discussing the very same strategy as mine.
But it's interesting how to price a strategy. One off is better for buyer if it works, better for seller if it doesn't. Also, share of profits is problematic: a seller need to know buyers' accounts size and need to prove when they trade it. Also, how can a buyer stop trading the strategy? Seller would like to keep receiving money forever, as there is no way to prove that buyer stopped trading it.
It is really annoying that some companies refuse to offer a demo. I understand the hacking issues but nowadays that doesn't make sense unless a software developer is totally unaware of modern software protection techniques.
I must admit I had trouble getting approved for a PAL demo. I was asked to provide proof of address and then answer a short questionnaire but I finally got through the process and I thought of applying it to Soybean Oil (BO) just to compare with the results attributed to TSL.
For the in_sample I choose daily data from 01/1970 to 12/31/1999. Since I am a swing trader I thought 1 full point for exit objective and the same for stop is good enough. This is R:R of 1 and I asked PAL to find patterns sufficiently profitable with profit factor greater than 1.30. The PAL output is submitted below:
Of course, we are not interested so much about the in_sample results about the out_of_sample. Just to make sure, I kept the out_of_sample data (01/2000 - present) on an external flash disk to make sure there was no secret snooping around taking place (I doubt it by the way). I will post the results with the next post.
Yes, this is why having someone who is proficient in structuring contracts would be imperative.
Personally, as a seller, I would structure the contract such that if the strategy continues to work, I get paid, but if the strategy stops working, the buyer stops paying me. Since I would be monitoring the strategy's results anyway, I would know if the buyer could conceivably have profited from its use. Whether they did or not would be irrelevant to me. I might even go so far as to force the buyer into setting up a trust specifically to trade the strategy as a segregated account, so that the results would be transparent to me and that would facilitate the profit-sharing aspects.
I continue now with out_of_sample, BO daily 01/02/2000 - 03/23/2012. Four patterns out of the 17 in the in_sample failed. The profit factor of the system of patterns in the out_of_sample is equal to 5.23. That is a fantastic profit factor for such a long out_of_sample. To me this may mean that PAL patterns have some predictive power. This is the out_of_sample output:
