Not really. Do you know how many times you can do that and still win? Not without a lot of statistical backtesting. What was described about averaging up/down is simply a lot of guesswork. There's no way to measure success or failure of such a strategy without coding it up and testing it.
And even with that, the strategy cannot take into account changing market conditions over the medium-term, which can change radically, as we have just seen over the past 3 years. Calm with turbulence, then calm seas, then absolute chaos.
And the probability of trading like that, even without extreme variables, is zero. Why? Because there's no probability distribution possible for future price. You don't know anything about future price movement in that scenario. The trader is simply guessing that by continuing to go against the market, that at some point it will reverse and he/she will walk out of a black hole. How do you calculate that risk?? Gravity will sneak up on you more often than not. The 2 point scenario is no different. You can't calculate the risk and it turns out to be a coin toss.
I'm agreeing with you here. You cannot backtest it and come up with a viable plan going forward, due to said changing market conditions. It means that backtesting is all BS. There is NO market that can provide you with a reliable backtest. None! NONE!
That's not true either. What Speedo described can be coded and tested. Variables can help include anomalies in the market. At the very least, the setup he described has a possibility. It's technical and the monkey wrenches can be worked out if found. He has numbers and numbers help. Hot air never helps.
Well I'll disagree to this extent: sure typical market patterns are usually unreliable because they've all been exploited. However, there are hidden patterns and correlations that, when found, can be exploited. Requires big data and computer power, but can be done. Think Markov and hidden Markov chains.
Well, I do not know stock, but if that "chain" helps you with future stock movements, fine. Makes the monies Lebowski!
Markov can be used to build trading models - here's a link to a short description. https://www.quantopian.com/posts/inferring-latent-states-using-a-gaussian-hidden-markov-model It's simply a "chain" of events where the events themselves may or may not be known but an outcome can be predicted in the current state. Speech recognition, computer learning are familiar areas which employ Markov chains. In any case, backtesting is still crucial to understand something about your idea. For instance, if you're trading the opening price under "x" conditions, you want to have a good understanding that "x" conditions show up enough times to be profitable. That's the essence of a backtest.
It is very depressing. I clicked your link, stared at the article for 30 min and had no clue what they were talking about. Then I googled Markov chain and read up in wiki, still have no clue and how to use it to trade.