We've defined the jump, shown the edge, written the rules and watched it play out in real companies, winners and losers alike. This last chapter is the one that actually defines how to build a portfolio: the rules to run it, what it costs, what the ride feels like, and whether you have the temperament to stay with it - as it's not easy. The strategy is mechanical, but staying with it is entirely human. So let me hand you the whole playbook.
The rules, as a weekly routine
This is a mechanical strategy, and it fits into about twenty minutes a week. Here is the entire method, start to finish.
Scan for StockRank jumps of +10 to +40 points that close the week at 90 or above - a genuine leap into the top decile, not a slow creep into the 80s.
Filter on size, and size alone. Keep only names roughly £10m to £350m in market cap. The £10m floor is a tradeability floor: below it you're in nano-cap territory where the bid-offer spread is so wide you could never realistically capture the edge. The £350m ceiling is where the neglect premium lives - above it, too many professionals are already watching.
Buy equal-weight, one slot of the book each. If more signals fire in a week than you have free slots, take the biggest jumps first and simply skip the rest.
Hold while the rank stays at 80 or above. Do nothing. This is most of the job.
Sell on the first weekly close below 80. One clean rule. Add a hard two-year long-stop for the rare name that drifts sideways at rank forever.
That's it. Notice what isn't there: no sector filter. The core strategy filters on size, full stop. The sector evidence from Chapter 5 - lean away from Financials and Energy, tilt toward Healthcare and Technology - is optional upside you can layer on top if you want it, although the sector rule may be hindsight bias.
Run it as a basket - around 20 names
This is not a one-stock strategy, and trying to use it as one is the surest way to be disappointed.
Remember the shape of the edge: means beat medians everywhere, because the returns live in a long right tail, a minority of big winners doing most of the lifting. That mathematics has a direct, practical consequence. You have to hold enough names to catch the tail.
A 20-slot basket is the sweet spot: +19.0% CAGR, the best Sharpe of any size at 1.32, and a maximum drawdown of -26%. A 10-stock version is liveable but lumpy - +17.3% CAGR with a similar -26% drawdown but a markedly lower Sharpe - and it skips about 66% of the signals that fire. That last number is the quiet killer: with only 10 slots, two investors running the identical rule can post wildly different years simply because of which signals they happened to have room for. Concentration doesn't just add risk here, it adds luck, and luck cuts both ways.
The risk reduction is the real gain. Holding 20 names rather than 5 cuts volatility by around 40% and shrinks the worst drawdown from -40% to -26% - for a CAGR that's barely different. Diversification isn't a hedge for ignorance; it's how you let the base rate do its job without a single bad name deciding your decade.
Free slots wait in cash
This strategy only buys when a qualifying jump appears. It is not a rule that keeps you fully invested at all times. When you sell a name on a rank break, that slot sits in cash until the next genuine signal fires. You don't reach down the market-cap scale or lower your standards to fill it.
In practice a 20-slot book runs at about 86% utilisation - a median of roughly 18 of 20 slots filled at any moment, with a couple waiting in cash for the next leap. That idle cash earns nothing in the model, a small drag that's already inside every return figure I've quoted you. And there's no rebalancing: winners are allowed to ride past their starting weight rather than being trimmed back. You let your Yü Groups run. The discipline is all on the sell side, governed by the rank, not by a rebalancing calendar like the NAPS or a target price.
What it costs to run
Every number in this study is net of our best estimate of trading costs. Run as a 20-slot book, you open about 22 positions a year and hold the median name for roughly 8 months, so the book turns over slightly more than once a year. That's active, but not frantic. The cost of each round trip depends on the size of the company - smaller means wider spreads:
Each round trip includes 0.5% stamp duty and 0.2% commissions, with the rest being the bid-offer spread you cross getting in and out. Add it all up across a year of turnover and the total cost drag is roughly 4% a year on the compound return. That 4% drag is the price of admission, and it's the honest reason to run a manageable ~20 names rather than churning a bigger, faster book. Trade less, keep more.
The shape of results
Take every +10 to +40 jump into the top decile, held to its rank-break exit - 853 events, all caps, price-only. Sort them worst to best. 53% made money. The mean return is +18.8%, but the median is just +1.5%. Read that gap again. The average outcome is nearly thirteen times the typical outcome, and the only thing that can produce a spread like that is a long right tail: a handful of enormous winners - the biggest being Sopheon at +1,150%, an 11.5-bagger - dragging the average above the middle of the pack.
This is the single most important thing to internalise. The basket does not make money because most of its stocks go up a bit. It makes money because a minority of its stocks go up a lot, and those few carry everyone else. The red left of that chart is the losers you will hold, counted honestly. The green right is the small band of multi-baggers that pay for them all. It's not about picking the winners in advance - you can't. It's about owning enough names, for long enough, that when a winner turns into a monster, you're holding it.
The hit rate you should expect
Hit rates across the study run 53% to 61%. So expect fewer than six positions in ten to be winners over a 12-month view, which means four or five in ten will be flat or losers. That is not a flaw in the strategy - it is the strategy. In a disciplined basket with the sell-below-80 exit, the winners ought to more than pay for the losers over time.
If a hit rate below two-in-three sounds low, that's exactly the point. The edge is the gap between your winners and your losers across the whole basket, not the success of any single selection. Chasing a higher strike rate by holding fewer, "surer" names throws away the tail - and the tail is where all the money is.
The temperament test
The numbers appear excellent, but the experience of earning them is uncomfortable, and of course the future may not be like the past. You'll buy a stock and watch it do nothing for months. You'll sell a position on a weekly close below 80 and watch it bounce the next week, feeling like an idiot. You'll hold a winner through a 20% drawdown while every instinct screams to bank the profit. You'll own names you've barely heard of, in sectors you don't follow, because the rank - not your opinion - put them there. And you'll sit on cash in a free slot while the market runs, wishing you were fully invested.
That's the temperament test, and it's not for everyone. The strategy requires trusting rules over your feelings, repeatedly, for years. Most people can't - which is likely why the edge persists. If everyone could hold the boring positions and take the painful exits, the neglect premium would be arbitraged away, as is more the case in the deeper US market. A mechanical rule only works if you're mechanical about following it.
So before you run this, ask yourself honestly: can I follow a rule I didn't feel like following? If the answer is yes, then the data is on your side. If it's no, well it's not for you.
Education, not share tips
This is not a recommendation, and this is not a list of stocks to buy. I've given you a tested, honest base rate and a mechanical rule for acting on it based on historic data - and that distinction is everything. Share tips make you dependent on experts. Rules on base-rates give us more agency. Share tips so often run out the moment the tipster goes silent - and often they only say buy, never sell. Rules based on base-rates you understand are empowering, and can be applied through time, to companies that don't even exist yet.
So this is our way of putting the odds a little more in your favour, and a framework for thinking clearly when the market is confusing. Invest with rules, not guesswork.
If you'd like the complete method - every confidence interval, the full cost model, the robustness checks, the US replication and the appendix of evidence - the full research study, The StockRanks Jump Effect, is available as a downloadable PDF. Read it and bring your questions back to the community. The best of these studies, after all, started with a community question!