If your algorithm works, try it using paper-trading. Maintain a log of the prices it would have bought at, and sold at, and what the overall profit would have been. The longer it runs and the more consistent it is at making a profit, the more it will appeal to investors.
Paper trading is a proxy for the real market, but no substitute. You will find you may not get filled as quickly as you would like or at as good a price. So if you believe in the algorithm, you could start with a little bit of money, even if after transaction costs it does not result in a profit. This will further improve its appeal to investors.
While selling subscriptions to the algorithm's picks might seam logical, there are a few problems with that. Investments have to be suitable - you don't want a retiree putting their life savings into the picks only to find there is a bad day and they lose it all, and they are bankrupt in retirement. Is it really worth having that on your consciences (and the resulting lawsuit) for the monthly subscription fee? Suitability is very important.
If you have a track record and have tested it, you may be able to find sophisticated investors (hedge funds, speculators etc.) who understand the risks and may be willing to invest in your algorithm if it fits their investment profile and they have capital to spare.
In all likelihood, unfortunately, the algorithm is something that only makes money some of the time and doesn't make money the rest of the time, or may work for a short time, but not in perpetuity. Or may make money up to a certain amount, but no more (due to market liquidity issues). It could also be exacerbated by the accounting methods used (there are easy ways to make mistakes or categorize losses as "long term investments").
The key is to find these issues early on, before a lot of money is involved, and a lot of people are upset.