Stock prediction service

[lucas.tychrom]

I have built an automated system that seems to be able to correctly predict upward stock movement with about 65% precision (i.e., whenever it makes a prediction, it is correct 65% of the time). The problem is that the prediction horizon is on the scale of minutes, so we're often pursuing price differentials on the scale of cents. Hence, fast and cost-efficient trading execution is just as important, and this is where I (as a retail trader) keep failing to make money due to price slippage, trades not getting executed, etc.

Instead of trying to solve the execution problems on my own, I'm considering providing a streaming service that a professional hedge fund with the right execution infrastructure and expertise would subscribe to, and pay me either a subscription fee or a percentage of profits resulting from use of my predictions.

Of course, there is no reason for a hedge fund to trust me, so I could provide a web API for accessing my service for free for one month. The web service would provide any predictions in real-time (i.e. "the symbol xx will go up over the next yy minutes"), and the hedge fund can easily use it to evaluate how accurate & useful my predictions are. After the free month, the hedge fund can decide whether to pay to keep using it or not.

I'm wondering what people's thoughts are on this?

 

[nonlinear5]

I have built an automated system that seems to be able to correctly predict upward stock movement with about 65% precision (i.e., whenever it makes a prediction, it is correct 65% of the time). The problem is that the prediction horizon is on the scale of minutes, so we're often pursuing price differentials on the scale of cents. Hence, fast and cost-efficient trading execution is just as important, and this is where I (as a retail trader) keep failing to make money due to price slippage, trades not getting executed, etc.

Instead of trying to solve the execution problems on my own, I'm considering providing a streaming service that a professional hedge fund with the right execution infrastructure and expertise would subscribe to, and pay me either a subscription fee or a percentage of profits resulting from use of my predictions.

Of course, there is no reason for a hedge fund to trust me, so I could provide a web API for accessing my service for free for one month. The web service would provide any predictions in real-time (i.e. "the symbol xx will go up over the next yy minutes"), and the hedge fund can easily use it to evaluate how accurate & useful my predictions are. After the free month, the hedge fund can decide whether to pay to keep using it or not.

I'm wondering what people's thoughts are on this?

1. "Percent correct" is a really bad performance metric. You can be correct 90% of the time, and still lose money. Vise versa, you can have a profitable system with only 25% win rate.

2. No one will take you seriously, unless you have an audited track record. Plug your web API to collective2.com, run it for 3 years, and if you have a decent performance, you may have subscribers.

 

[algofy]

I only buy stock prediction services with 70% precision or better, do you have a slightly better system for sale? PM me

 

[lucas.tychrom]

1. "Percent correct" is a really bad performance metric. You can be correct 90% of the time, and still lose money. Vise versa, you can have a profitable system with only 25% win rate.

2. No one will take you seriously, unless you have an audited track record. Plug your web API to collective2.com, run it for 3 years, and if you have a decent performance, you may have subscribers.

1. That is correct. I should add that the profit-to-loss ratios are also pretty good (>2) as long as my trades execute without significant slippage.

2. That's my problem: all I can be confident of right now is my predictions of the stock price movement, not my trades. To get a track record, I need to make real trades, but right now I'm not able to make much with retail trading. It doesn't matter whether I use collective2.com or directly interactive brokers, etc., right? I can build a trading record either way.

 

[nonlinear5]

To get a track record, I need to make real trades, but right now I'm not able to make much with retail trading. It doesn't matter whether I use collective2.com or directly interactive brokers, etc., right?

With collective2, you don't have to execute trades. You can just generate a signal to buy/sell, and collective2 simulates trades based on these signals.

 

[eliteubertrader]

collective2 became really expensive

 

[Jack1960]

Minutes us too slow. It is not 1980. You need milliseconds if you want to sell it.

 

[wrbtrader]

...I'm wondering what people's thoughts are on this?

How well connected are you to the hedge fund industry ?

Do you currently know those at hedge fund firms that makes purchasing decisions ?

How will you differentiate an individual trading their own capital from a hedge fund when they purchase your automated trading system ?

I only know a few in the hedge fund business but I do know they will not do business with someone that's also marketing their system to other hedge funds (the competition). Therefore, most likely you'll need to be a lot more personal (as in person) than building a network with hedge funds via a web API service.

 

[Pekelo]

collective2 became really expensive

You mean running your own business for $3-400 annually is expensive? Try buying a cab or hot dog vendor license in NYC...

But just to be on topic, C2 would be the best option for OP...

 

[MarkBrown]

i have a model that doesn't lose ever with a real track record - ok shutting up now have to get back to shopping, a daunting task let me tell ya.