Papers With Backtest

Switching from RSS polling to SSE news streams shaves seconds off your latency. The real question nobody backtests: by the time 'breaking' hits your feed, market makers have already repriced. Faster pipes don't create alpha if the signal decayed before you saw it.

@Greenbackd Sign matters: dealer hedging amplifies in negative gamma but dampens in positive gamma

Flows hedging flows hedging flows. The sheer scale of call option activity means the tail wags the dog. Options are now driving the underlying market. Dealer hedging increasingly determine short-term price action. Reflexivity has become structurally embedded in equities.

"It's too quiet" is one of the most expensive instincts in markets. Long-vol in low-VIX regimes bleeds carry far more reliably than it pays off in spikes. The edge isn't predicting the breakout. It's pricing the carry while you wait.

Oldies but Goodies Stocks with high accruals underperform. Split accruals into industry vs firm-specific: only the firm-specific part predicts returns—mispricing, not risk. paperswithbacktest.com/strategies/acc…

Pairs trading scales badly not because the signal dies, but because two legs eat capital. The real evolution is portfolio stat arb — one leg against a basket. Most retail quants optimize the signal when the binding constraint is buying power.

Markets being efficient and L/S specialists making money aren't contradictions. Large-cap alpha exists — it's just thin, slow, and expensive to harvest. The edge isn't a sharper view of fair value. It's signal combination, sizing, and tolerance for being right too early.

Pre-tax Sharpe is fiction for taxable accounts. After-tax, after-cost compounding is the only metric that matters. Most 'tax vs concentration' debates evaporate once you backtest the actual drag — and the clever structures rarely win.

Stacking uncorrelated carry trades is the closest thing to a free lunch: return scales linearly, vol scales as sqrt. The catch: most 'diversified carry' books are one factor in five costumes. Test the combined Sharpe net of costs, not whether the legs look different.

Most trend traders argue about entry rules. The signal is the cheapest part of the system. Sizing across speeds, exits when the regime flips, and costs netted across asset classes are where the Sharpe actually lives.

Buy 10 units by month-end, benchmark is the avg of the last 5 closes. Optimal replication: 2 units at each of those 5 closes, exactly. Any other schedule is a directional bet dressed as execution. Most execution alpha against a VWAP-style benchmark is hidden beta.

Oldies but Goodies Combine multiple return forecasts via goal programming: pick weights that maximize mean and skewness while minimizing variance. Tested monthly on S&P 500, FTSE 100, and Nikkei 225 with a long/short index vs T-bill rule. paperswithbacktest.com/strategies/usi…

Every regime-change pitch frames active as the answer. The data says otherwise: SPIVA shows the majority of active funds underperform in bull, bear, and sideways markets. The constant isn't the regime. It's the fee drag.

A Nasdaq microcap dropping 87% in a day isn't a scam story — it's a capacity story. Backtests assume you can exit at the print. The print is often a WhatsApp pump. In microcaps, liquidity quietly kills every theoretical edge.

@JoeTigay Vol-based hedges preserve upside but bleed skew premium in calm regimes. AQR pegged systematic put-buying drag near 3-4%/yr outside crises.

Most hedged equity funds offer "soft protection" that fails on the way down and caps you on the way up. We spent April proving why an open ceiling and volatility-based hedging is the better way to play market rips. open.substack.com/pub/joetigay/p…

Swapping LIBOR for SOFR in old pricing models isn't find-and-replace. LIBOR was unsecured, forward-looking, term. SOFR is secured, backward-looking, overnight. The basis isn't noise — it's bank credit risk made explicit. Theory is cheap; conventions kill you.

Oldies but Goodies Country demographics predict equity returns. Each year, sort countries by lagged growth in average age and buy the fastest-aging quartile to capture a risk premium. paperswithbacktest.com/strategies/dem…

VIX isn't a fear gauge. It's a 30-day forward implied vol of SPX options, which is why expiry sits 30 days before SPX OPEX. The tradable signal isn't the level. It's the basis between VIX cash and front-month futures.

@quantian1 Daily rebal of 3x ETFs is mechanically pro-momentum: every up day forces buys into the close. Once that flow rivals real demand, the gamma sets the marginal print. The loop breaks when inflows stall - and then it unwinds the other way, hard.

I think that if you are wondering why a theme is up 35 of the last 40 days, and the 3x levered ETF on that theme routinely does $10bn+ of volume and is consistently ranks #3 behind SPY and QQQ as the most active ETF, there might be a simpler answer than "we are still so early".

New quants think the bottleneck is alpha creativity. It rarely is. The compounding skills are elsewhere: clean data, honest backtests with costs, knowing when a Sharpe is noise. Most alpha is already known. The edge is in the discipline of testing it.

Momentum 'struggling' in high vol isn't a bug, it's the premium. The factor pays you for sitting through volatility-driven crashes. A momentum backtest with no 2009-style drawdown isn't momentum — it's overfitting in disguise.