What a strategy really is
"I like solid companies" is an opinion. Nobody can test it, and on a bad Tuesday it will not tell you what to do. A strategy is the same instinct turned into decisions specific enough that a computer could follow them without asking you anything.
There are four of those decisions, and every plan makes them whether or not its owner realises it. What will I buy. When do I buy it. When do I sell it. How much of the account can any single position take.
People who skip these do not avoid making the decisions. They make them by accident, at the worst possible moment, based on how they feel that day.
Decision one: what you will buy
Start with a group of companies rather than individual hunches. You might begin with companies you know and use, and let the tool find similar ones by financial health: those that make real profit, do not carry crushing debt, and are large enough that a single bad quarter will not end them.
The value of describing the group rather than naming stocks is that the group survives when individual companies do not. If one falls out of favour, the plan replaces it with another that meets the same description, and you never have to make that call under pressure.
A reasonable first group might be twenty to eighty companies that pass your filters, from which the plan holds a handful at a time. Too narrow and one company dominates; too broad and you own a slightly worse version of the whole market.
Decision two and three: when you buy and when you sell
Buying is usually the easy half: on review day, buy the companies that meet your rules and are not already held. What matters is how often review day comes around. Monthly is a sensible default; weekly trades more and pays more fees; yearly is slow to respond when a company deteriorates.
Selling is where plans live or die, and it needs to be decided in advance because it is impossible to decide well in the moment. Three sensible triggers: the company no longer meets the rules that got it bought, it has fallen past a loss level you chose, or it has grown so large in the account that it needs trimming.
Write these as numbers, not feelings. "Sells a stock once its loss reaches 10%" is a rule. "Sell if it looks bad" is an invitation to panic.
Decision four: how much any one stock may take
If you hold five stocks in equal amounts, each is 20% of the account, and a company that halves costs you 10% of everything. If one of them has run up to 45% of the account, that same halving costs you 22%.
So set a cap. A rule such as "no stock may take more than a quarter of the account" means that when a winner grows past that line, the plan sells part of it and redistributes. This feels wrong the first time it happens, because you are selling something that is working.
It is worth understanding why the rule exists anyway. It is not there to maximise your best case, it is there to make sure your worst case does not come from one company you happened to be overweight in when the news broke.
A worked plan, start to finish
Buy: companies that made a profit in each of the last four quarters, carry manageable debt, and are worth more than $10 billion. Hold eight of them at a time.
Review: once a month. Anything that no longer meets the rules is sold and replaced by the best available company that does.
Sell early: any stock that falls 10% below what was paid for it goes immediately, without waiting for review day. Cap: no stock may exceed 25% of the account; anything larger is trimmed back at review.
That plan is fully specified. A computer can run it, a history test can check it, and you can read it back in a sentence. Whether it is any good is exactly what the test is for.
Testing it and changing one thing at a time
Run it across five or more years and look at three things: how the account ended, how far it fell at its worst, and how many trades it took. Then change exactly one rule and run it again.
One rule at a time is the entire discipline. Change the review from monthly to weekly and you learn what review frequency costs in fees and gains in responsiveness. Change three things at once and you learn nothing, because you cannot tell which change did what.
Resist the urge to keep tweaking until the result looks wonderful. Past a certain point you are no longer improving a plan, you are fitting it to history, and a plan fitted to the past tends to disappoint in the future.
Building one in Stax
You can describe all four decisions in conversation rather than filling in a settings page. Buddi asks what matters to you, checks the companies you mention, and writes the rules in plain words for you to approve or argue with.
The rules that come back are readable: which stocks it holds, when it reviews them, the exact loss at which it sells, and the largest share any one stock may take. If a rule is not one you would defend out loud, change it before you test it.