I seldom do paper trading. Reasons: 1. My strategy doesn't cause a market impact. 2. My strategy is interday trading, not intraday. I mainly trade at market close. Besides, paper trading is sometimes uselss. Imagine the current price of stock A is $100. Best Bid .............. Best Ask $99.99 .............. $100.02 If you use a paper trading account and submit a traditional market order to buy, your buy execution price will be $100. If you use real money to trade, the execution price will be $100.02. That's why paper trading is sometimes useless. But yes, if your strategy needs to be tested by paper trading, you have to. However, for me who only trade liquid ETFs and trade at market close, I don't need to do paper trading.
Backtesting is what I rely on. Continuous data flow is needed for testing the true viability of the algorithm, I use past data, depth, volume, price. And any new program or indicator I first use on small trades, I have an account with fxview exclusively for this, as there are no deposits that need to be maintained and there are no inactivity charges so it’s a good way to see the efficacy on live environment. If algo seems solid, I take it live on higher capital with IB (which is more like my main account).
This book is one of the best you will find. THere are going very deep in to optimization and what ist optimization and how do you do this correct. @KevinDavey I dont like him, his books are totaly waste of time and money. 1 or 2 Years after his first book i ask him on future.io (bigmike) how they doing this money/random test and he say , that he dont do this anymore. There for in his "Curse" he optimization 3 paramter at once on walk-forward (and walkforward use the best! paramter....not the stable one..) In the videos he not showing the DD or what a "step" mean. change from 5-10 or 5-100 ?? Better: bettertraderacademy / Tomas Nesnidal or Andrea Unger