Exactly my point, I guess the general definition nowadays is if someone uses a moving average that their charting package has to calculate the value of they are "Algo Traders".
Losing and more than a year or two. Retail who only know how to program and not trade seldom ever make it cause you have to learn how to make money first, you have to continually add to your mental library every year and once you get one system automated, it is never ending of tweaks until you have traded through a few cycles or decades and recorded thousands of different climates until you have seen it all a few times. Will never see 911 cycle or Swiss France due to brexit until another one happens, but you bank the occasion and then solve so you have options on the next time and code it. I never think about how much can I make, I walked away 8 years ago on making the most to how can I seldom lose and what is it going to take to do this, this is much more of a challenge as you will get many who say something can't be done which is more of encouragement to keep this path, often times people want you to fail cause THEY can't accomplish it, you have to walk alone closer you get to your goals. And gets very lonely cause you can't trust anyone as all can be more to of enemy cause all are trying to get better prices. Trading like a boxing match, in the ring you hate guy you fighting but you may respect them outside of ring. Automation is what you put into it and there is much math in it even if just using trendlines, "NOTHING" is exact in the world, well maybe space vehicles, but there are small differences of patterns like ascending triangle, does the highs all have to be exact tick or can it be higher/lower by couple ticks, so your back testing comes into play. Before I get to stage of actually coding for automation, go through many process first to make you have edges first. I am very conservative so it might take 18-24 months from start of an idea to it real time trading. And never stop doing some type of manual trading, you will really lose it if you stop. It comes down to your account size and only after you know your app is profitable, you never want to go more than a percentage of the account into the markets, anything can and will happen that is NOT in your favor at least once or fifteen times or more, very hard lessons but learn how to overcome before they happen, GoT to have the answers first before question and program it in. And you can hedge scalping or day trading but not in conventional ways/methods, and yes it costs to do so, but don't want to lose thousands cause something blow up. It like fire insurance on your house, you hope you never have to use it. When you start with small account you can make nice size percentages, as account gets bigger, numbers affect one's thinking. Good luck to you.
As others have stated, impossible to give an answer as it depends on different factors. I believe any trader regardless of style should be able as a minimum ballpark earn a annual 20% on equity/capital. If not they are wasting time and should just do long term investing instead of trading.
Properly backtest, realize that 99.99% of the stuff you can think of will not work in practice (thus you don't need to lose any money trying it). Trading your algo live on e.g. IB's practice account is a good idea - sorting out the differences in your backtest and forward simulated performance this way is key to at least not losing money. If you're a lazy dreamer who hasn't tested properly you will lose or accidentally win because your sloppy algo is long and the market is in its bull phase.
Just IMHO: Sharpe 0.5 may be exceeded by monkies selecting stocks (it was the repeatable case in a Swedish zoo). Sharpe 1 is tradable IMO since it's less likely to be luck over a long sample period. Sharpe 2 is a decent. Sharpe 3+ is where getting rich is at. Tongue-in-cheek regarding the monkey; for a sufficiently small number of trades/sample period you can get any Sharpe that does not hold long term...
Okaaaaaaaay, you have a degree in financial engineering and you have asked the following questions; How much can a retail algo trader expect to earn? What should my expectations be for the year or two? What can be my sharpe, if i get lucky? Most successful traders are expert at identifying patterns on either the micro or macro level, or both. Your professed degree and your questions do not fit a pattern on any level.
Sharpe of 3+ is where getting rich QUICK is at. Sharpe of 2 will get you there but a little bit slower. But a Sharpe or 1 or below will take ages
Is there a reason why you pegging 20% ROE and not any other number? Anyone who can reliably produce 7-10% ROC with a Sharpe of 0.7-1 is doing OK, IMHO, especially if they are completely de-correlated from the main risk factors (equity and bonds).