Desbloquear esta nota

0050 rose more than tenfold in ten years. Is picking your own stocks still worth it? We took the four factors you hear about most, held just five names a month, and backtested it properly. This Taiwan stock strategy (Taiwan Four-Factor Top 5, official and free) does one thing each month — after monthly revenue reports come out, it ranks every currently listed Taiwan stock on revenue momentum, price momentum, ROE and low volatility, and holds the five with the best combined score, trading at the next day's open. Backtested January 2016 to October 2026: +2,384% total return, about 34.8% annualized, Sharpe 1.45, max drawdown −37.9%. It beat simply holding the 0050 ETF in 8 of the 10 full calendar years from 2016 to 2025. How are the four factors scored? Revenue momentum, price momentum, ROE, low volatility Each factor is ranked as a percentile across the market, and the four percentiles are added up: First, revenue momentum: average monthly revenue over the last 3 months divided by the 12-month average — is the business picking up? Second, price momentum: the return over the last 120 trading days. Third, ROE: trailing four-quarter return on equity — how well the company turns capital into profit. Fourth, low volatility: the standard deviation of daily returns over 60 days; steadier stocks score higher. A stock needs all four numbers to be ranked. ETFs and financial/insurance stocks are excluded, as are illiquid names averaging under NT$30 million a day in trading value over 20 days, and any stock under disposition (Taiwan's trading-restriction list). The top five are weighted by score squared; in practice their scores sit so close together that the weights come out nearly equal. Why wait for monthly revenue and rebalance after the 12th? Taiwan-listed companies report monthly revenue by the 10th. The strategy waits for the first trading day on or after the 12th of each month, scores at the close and trades at the next open, so every number it uses was already public. ROE works the same way: only figures released before the trading day count, never numbers that weren't known yet. If a holding is put under disposition, it is sold at the next open. Backtest performance (2016–2026, daily bars) MetricValueTotal return+2,384% (0050 +945%)Annualized~34.8% (0050 ~24.4%)Sharpe1.45Max drawdown−37.9% (0050 −34.0%)Random-pick benchmarkAgainst 1,000 random portfolios just as concentrated: none had a higher Sharpe, and it out-returned 99.8% Includes 0.3% cost per side: 0.1425% brokerage each way (no broker discount) plus the 0.3% transaction tax on sells, averaged out to about 0.3% per side. Backtest results, not a guarantee of future performance. Five things to know before you run it First, the stock list is today's listed companies. Stocks that were delisted or acquired over the decade aren't in it, and that survivorship bias flatters the backtest — expect live results to come in lower. Second, five names means deeper drawdowns: from November 2021 to October 2022 it fell 37.9%, more than 0050, and didn't reclaim its high until June 2023. Third, it lags when a few mega-caps carry the market: it trailed 0050 in 2016 (+20.2% vs +27.2%) and 2020 (+10.9% vs +31.1%), and is well behind so far in 2026 (+33.9% vs +69.5%) — when TSMC does the heavy lifting, a five-stock basket without it can't keep up. Fourth, there is no out-of-sample period: the parameters were set on the same data. Shifting the momentum and volatility windows two steps either way gives a neighborhood-average Sharpe of 0.92× the chosen setting — a plateau, not a lucky spike — but that is still no promise about the future. Fifth, fill prices are assumed: the backtest assumes fills at the next open; opening auctions can fill at worse prices, and slippage isn't included in the cost. The rules are clear. Running them on time every month is the work. The logic is all above. What takes time is pulling revenue, financials and prices for the whole market every month, computing four factors, ranking, dropping restricted and illiquid names, and rebalancing on schedule without missing a month. Blave Agent runs that chain for you: add this free strategy from the library, then tell your Agent to deploy it. See the Taiwan Four-Factor strategy in the library →