Building CetaResearch. Data analysis for policy, research, and startups.
value investing
We backtested Graham's net-net strategy on Japan's JPX exchange from 2006 to 2024. Japan net-nets returned 7.39% CAGR vs 4.32% for the Nikkei 225, a +3.1% annual excess with 45% down capture. Japan is the major developed market where the net-net premium shows up most clearly in the data.
backtests
The Graham Number screen applied to Canadian stocks (TSX) returned 10.89% annually from 2000–2024, beating the TSX Composite by 6.46% per year. Down capture: 51.8%. Up capture: 159.5%. In years when the TSX fell, this portfolio fell only about half as much.
value investing
We screened NYSE, NASDAQ, and AMEX for D/E < 0.5 and Piotroski F-Score >= 7, then held the portfolio annually for 25 years. The result: 7.57% CAGR vs 7.85% for SPY, with max drawdown of -31.41% vs -38.01%. Lower return, but meaningfully better downside protection.
momentum
We tested the 52-week high proximity strategy on Korean stocks (KSC) from 2000 to 2025. 12.07% CAGR, +7.26% vs KOSPI, 37.9% down capture. 2024 returned +25.1% when KOSPI fell 10.2%.
revenue acceleration
We ran revenue acceleration on all US stocks from 2000 to 2025. 3.56% CAGR vs 7.33% for the S&P 500, 107% down capture, fully invested every year. Every year from 2011 on lost to the index.
backtests
India produced the highest CAGR of any exchange in our 13-market sector momentum study: 21.45% annually over 26 years, +10.21% excess vs Sensex, 159.36% up capture. The drawdown is severe at -66.28%, and 120.56% down capture means it amplifies bear markets.
P/E mean reversion
We tested a sector-relative P/E mean reversion screen on 25 years of US stock data. Buy quality stocks trading 40%+ below their sector's median P/E. The strategy returned 10.25% annually vs 7.64% for SPY, with 2.61% excess return and zero cash periods across 25 years.
ai
Use your existing AI subscription to query 163 tables of financial data on Ceta Research. No API key, no setup. Copy the schema, paste into your AI, and ask questions in plain English.
ROIC
Sustained high ROIC on LSE stocks from 2000 to 2025. 7.62% CAGR vs 7.30% SPY. Borderline alpha (+0.32%), but the quality spread is real: sustained beats low ROIC by 2.74% annually. Down capture of 74.9%.
NCAV
Graham net-net backtest on Hong Kong stocks (2001-2024). -0.26% CAGR vs Hang Seng's +2.53% (-2.80% deficit). Pre-2015: powerful recoveries. Post-2015: structural collapse alongside Hong Kong's market decline.
NCAV
Xiao and Arnold (2008) found 19.7%/yr for UK net-nets 1980-2005. Our 2001-2024 backtest: +0.77% CAGR vs FTSE +1.81% (-1.04% deficit). Strong recovery trades (+78% in 2020) but struggles in normal markets.
NCAV
NCAV screen on Korean stocks from 2001 to 2024. 6.60% CAGR with only 22% volatility. Max drawdown -39.3%, lower than SPY. The Korea discount creates a reliable universe of balance sheet bargains.