[Masters of Investing] The master who preached holding winners, but his own rules actually make you sell in just a week

There's one stock market saying you hear all the time. Cut your losses short and ride your winners long. It's in every book, on every YouTube channel. I believed it too for years.

Here's the strange part though—half of it holds up to scrutiny. Plenty of people have run the numbers on cutting losses short, but there's surprisingly little hard data on whether holding winners longer actually makes you money. That's the part people rarely verify with actual numbers.

So I pulled out Christian Koulmavis's rules. He's known for buying breakouts and selling as he trails a moving average. I went through 12 years of Korean stock data for every signal that matched his entry conditions—came up with 3,880 setups—then applied his exit rules exactly as stated.

What surprised me most about the results wasn't the returns. It was the holding periods.

I matched his entries exactly

Throwing this rule at any stock is waste, not validation. But Koulmavis published his criteria. There had to be a sharp spike of 30% to over 100% in the last one or two months, a subsequent tightening into consolidation, and then a breakout above that consolidation.

I coded it exactly. Sorted by shortest moving average first, filtered to only stocks trading over 10 billion won daily so there's real liquidity. Entry is the close on the breakout day. I used only information available by market close that day.

Up to this point, we're identical. The only difference is the exit rule.

One holds for 20 days then sells. The other waits for the first close below the 10-day average then sells. This is a conceptual diagram, not actual data.

Here's his exact wording. Trail your remaining shares along the 10-day or 20-day line, and wait for the first close below that line as your exit. Your stop is the low of your entry day. I tested the 10-day version, the one recommended for beginners.

When I ran this across those 3,880 setups, I got numbers I wasn't expecting.

It wasn't a rule for holding winners long

Average holding time was one week. Half were closed out in six days. Only 1 in 33 trades lasted over 20 days.

This is supposed to be the poster child for holding winners, but what I found was it almost always takes you out early. The reason is simple: after a breakout, it's rare for a stock to never pull back for over 10 days. Any pullback touches that 10-day line.

Did selling this early actually improve performance? If you just look at average returns, the answer's underwhelming. Hold for 20 days: +0.22% per trade. Sell early: +0.33%. Basically identical.

So no difference if the averages match? Not so fast. The real difference shows up at the extremes.

What got cut was losses, not winners

Look at the worst decile: the 20-day hold averaged over 16% losses while the early exit version stopped around 12%. The 20-day hold had four times as many trades that lost over 20%.

Left is losses, right is gains. The red bars moved inward on both sides.

But the right side got clipped too. Winners in the top decile dropped from the 40% range down to 32%, and the share of trades over 30% gains fell from 5.5% to 3.6%. Exit early and you miss the big ones.

But these two things have completely different personalities. One repeated every single year. One didn't.

Red bars are lower in 11 out of 12 years. The one remaining year had zero trades over 20% loss on both sides.

The effect of cutting large losses never failed once in 12 years. Didn't matter if it was a bull market or bear market. When I twisted the entry conditions eight different ways and re-ran it, 15 out of 16 variations showed the same pattern.

Which brings me back to the original question: what if you actually do hold longer?

The longer-hold version looked good on average

Korean sources often mention a looser version: hold until the 10-day crosses below the 20-day. On the same setups, average holding time stretches to 19 days and the average return jumps to +0.85%—best of the three.

On paper, it looks like holding longer pays off. But break it down by year and that edge only showed up in 5 out of 12 years. When I tested 18 different condition combinations, not a single one consistently reproduced the outperformance across years.

Here's why the average was higher: take out the top 5% of winners and this version becomes the worst of the three. One moonshot in 20 trades was covering the losses on the other 19. And since those rare blowups cluster in certain years, you can't predict when they're coming.

That's my interpretation, but here's what's certain: the edge from holding longer never reliably came back year after year, while the edge from cutting larger losses showed up consistently across all 12 years.

So what's the answer?

These days I read 'hold your winners' differently. The real tool the master was actually using was simply a predetermined line for when to exit. That line wasn't there to maximize gains—it was there to tell him quickly when he was wrong.

Holding long isn't a skill by itself. What 12 years of data showed is that extending holding periods didn't deliver stability—it delivered bigger swings.

The one thing that never failed across 12 years was cutting the size of your losses. When we looked at stop losses in this series earlier, we got the exact same answer in the same place.

If you use the long-hold approach, you need the bankroll and the stomach to survive all the losses until that one moonshot arrives. Without both, you're just a strategy that loses 19 times to win once. I've watched plenty of people break here.

One sentence: The only part of the master's rules that survived 12 years was knowing when to quit—not how long to hold.

This is a record of historical price data and doesn't recommend trading any specific stock. All investment decisions and their consequences are yours alone.