OVVO Labs

Thank you for the cheap $COHR puts today!

🚨 New Options Feature Alert 🚨 We've had success trading options positions that combine insights from both of our tools. We previously shared how our options tool separates put and call IVs, and how the resulting ratio identifies conditional tail geometries. 👉 lnkd.in/p/gaKG--ik We also demonstrated our expected partial-moments ranking tool and its effectiveness. 👉linkedin.com/feed/update/urn… Using both, we formalized the strategy and ran it out of sample for the first time this quarter. Results, June 1 to September 18, 2026, across 60 tickers and 1,109 debit spreads (bear put + bull call): * 71.1% win rate * 41.7% return on capital just this quarter, not annualized! Bear put spreads on bearish-ranked names outperformed bull call spreads on bullish-ranked names by a wide margin: 48.5% ROC and 72.6% win rate vs. 28.7% ROC and 68.4% win rate. Notable given the market has spent most of the window near highs. The IV ratio flags when tail risk is asymmetric, the ranking tool flags which names. The strategy is what happens when you combine them. If interested in learning more, find us and send a DM through LinkedIn (no anons). • Past performance is not indicative of future results • Not investment advice • Not an offer to sell or a solicitation to buy securities

So there’s been a lot of chatter about portable alpha. The standard framing is simple: Keep your beta allocation. Then overlay a separate alpha engine. In practice, that alpha engine is often a CTA, managed futures, trend-following strategy, or some other systematic directional program. And what is that, really? Market timing. Systematic market timing, yes. Rules-based, diversified, liquid, and mechanically executed. But still market timing. CTAs enter, exit, and size positions based on price action across equity indices, rates, currencies, and commodities. The industry has been comfortable calling that portable alpha for decades because the process is systematic rather than discretionary. So the real issue is not market timing. The real issue is whether the timing rule is disciplined, liquid, scalable, and checkable. That raises a more interesting question: If allocators are comfortable using systematic timing as an alpha overlay, why should they be uncomfortable using systematic timing to manage beta directly? Most portfolios treat beta as static. You own the S&P 500, so your beta is approximately 1. You lever it, your beta is greater than 1. You hedge it, your beta is lower. You short it, your beta is negative. But why should beta be fixed? If market structure changes continuously, then exposure to that market should be allowed to change as well. That is the role of the OVVO scalar. The scalar is a rules-based exposure value derived from observed market structure. It determines whether the portfolio should be long, short, neutral, defensive, or aggressive — all through a single continuously updated beta decision. This is different from a traditional CTA overlay. A CTA is usually added beside the core beta book for diversification. Direct beta management changes the exposure of the beta instrument itself. That distinction matters. But the intellectual standard should be the same: Do not ask whether it is timing. It is. Ask whether the timing process is systematic, transparent enough to evaluate, liquid enough to implement, and checkable against actual execution. Portable alpha asks: “What alpha stream can I add to my beta book?” Direct beta management asks: “What should my beta be right now?” Static beta is an assumption. Dynamic beta is a decision. The OVVO scalar is that decision rule. More on the mechanism ovvolabs.com/media

🤯 Knew each of the component tools were performant (Ranking and the IV-ratio), but did not know the synthesis would work this well! 65 settled days. 40.6% mean ROC. 98.5% of those days green. Worst day −17.6%. Best day 82%. ovvolabs.com Not advice. Past quarter is not the next one. Tickers used this quarter: SMCI COHR VRT CVNA TTD TPL MU AMAT EXE CIEN HOOD PLTR VST LRCX ADBE KLAC WDC TSLA NEM EL TER ARES LITE IRM STX APP MRNA CEG APTV GLW DASH MPWR NOW SW WDAY NRG EME APH FCX KKR MCHP DECK FIX MOS SWKS GEV CCL TECH ON CRWD RDDT DXCM MRVL NVDA FSLR ALB NCLH AXON ANET APA

Missing data is a common problem in almost every dataset, and it can seriously affect the validity of analyses if not handled properly. A widely used solution is missing data imputation, where the gaps in the data are estimated and replaced with plausible values rather than simply discarding incomplete observations. There are many different missing data imputation methods available, and it’s often difficult to decide which method is best suited for a given dataset. A method I recently came across, which shows promising results, is NNS imputation. This approach is based on non-linear, non-parametric regression and estimates missing values by identifying locally optimal relationships in the data. It adapts well to complex patterns, including non-linear trends and varying variance, without assuming linearity or homoscedasticity. The image below illustrates two applications of NNS imputation. On the left, it is applied to a dataset with a non-linear relationship and 15% missing values. On the right, it is used on a heteroscedastic dataset with 30% missing values. In both cases, the imputed values (shown in magenta) closely follow the structure of the observed data (shown in black), showing that the method can recover underlying patterns even in challenging scenarios. NNS imputation can be implemented in R through the NNS package, which makes it straightforward to apply this method to real-world datasets. Thanks to Fred Viole for creating this awesome package: github.com/OVVO-Financial/NN… Unfortunately, I haven’t seen much literature comparing NNS imputation with other well-established imputation techniques such as predictive mean matching or random forest imputation. I’d be very interested to hear from your experiences: have you used NNS imputation yourself, and how did it perform compared to these other approaches? If you want to learn more about R programming, Python, statistics, and data science, make sure to check out my newsletter. Learn more by visiting this link: eepurl.com/gH6myT #RStats #Data #programming #rstudioglobal #statisticians

So there’s been a lot of chatter about portable alpha. The standard framing is simple: Keep your beta allocation. Then overlay a separate alpha engine. In practice, that alpha engine is often a CTA, managed futures, trend-following strategy, or some other systematic directional program. And what is that, really? Market timing. Systematic market timing, yes. Rules-based, diversified, liquid, and mechanically executed. But still market timing. CTAs enter, exit, and size positions based on price action across equity indices, rates, currencies, and commodities. The industry has been comfortable calling that portable alpha for decades because the process is systematic rather than discretionary. So the real issue is not market timing. The real issue is whether the timing rule is disciplined, liquid, scalable, and checkable. That raises a more interesting question: If allocators are comfortable using systematic timing as an alpha overlay, why should they be uncomfortable using systematic timing to manage beta directly? Most portfolios treat beta as static. You own the S&P 500, so your beta is approximately 1. You lever it, your beta is greater than 1. You hedge it, your beta is lower. You short it, your beta is negative. But why should beta be fixed? If market structure changes continuously, then exposure to that market should be allowed to change as well. That is the role of the OVVO scalar. The scalar is a rules-based exposure value derived from observed market structure. It determines whether the portfolio should be long, short, neutral, defensive, or aggressive — all through a single continuously updated beta decision. This is different from a traditional CTA overlay. A CTA is usually added beside the core beta book for diversification. Direct beta management changes the exposure of the beta instrument itself. That distinction matters. But the intellectual standard should be the same: Do not ask whether it is timing. It is. Ask whether the timing process is systematic, transparent enough to evaluate, liquid enough to implement, and checkable against actual execution. Portable alpha asks: “What alpha stream can I add to my beta book?” Direct beta management asks: “What should my beta be right now?” Static beta is an assumption. Dynamic beta is a decision. The OVVO scalar is that decision rule. More on the mechanism ovvolabs.com/media