The information lag in spirits data
IWS R (formerly IWSR), Nielsen, and Circana all publish spirits market data. It's useful data. It's also 90-180 days old by the time it reaches most brand decision-makers, filtered through distribution layers that can obscure regional nuance, and expensive enough that most independent operators and small brands can't access it.
By the time a small brand learns from published data that a competing category is gaining share in their target market, the buying decisions that would have been affected by that information have already been made.
Forward-looking information, about what's likely to happen rather than what happened, has a different value curve. If you know three months in advance that buyers in your key markets are expecting a specific category to grow, you can adjust your allocation, your sales rep emphasis, and your trade marketing before the data confirms what you already knew.
How the prediction market surfaces what sales data misses
The participants in Liquor Bets are, by self-selection, people who are paying close attention to the spirits industry. Buyers, brand managers, distributors, bartenders, enthusiasts, journalists, consultants. Each person brings a different vantage point.
The bartender in a high-volume cocktail bar in Nashville sees which bottles are moving in the well. The buyer for a 12-location retailer in Texas sees what's sitting versus what's turning. The brand manager at a craft distillery sees their own sell-through data and their conversations with distribution partners.
None of these people have complete information. But together, their predictions, weighted by conviction and aggregated across all participants, produce a signal that's faster and more granular than anything in published trade data.
Specific examples from recent Liquor Bets markets:
We ran a market on whether Japanese whisky would maintain its current premium positioning through 2026 against growing supply. The market came in at 58% probability of maintaining. Several participants who had distribution relationships in the market noted that secondary positioning (Japanese whisky in more accessible price tiers) was already happening in their accounts. This was visible in the comments and reasoning participants attached to their predictions, months before it would have shown up in category data.
We ran a market on whether any of the three major Caribbean rum brands currently in rebrand processes would break through to the premium on-premise tier meaningfully by year-end. Market came in at 31% probability. The low probability reflected informed skepticism about whether the trade marketing investment behind those rebrands was sufficient to shift bartender perception, which is the actual gating factor for premium on-premise adoption. Sales data wouldn't have shown this until after the fact.
The brand intelligence layer
Beyond category-level questions, prediction markets can surface something close to real-time brand perception tracking. Not sentiment analysis of social media (which is a different and noisier signal). Actual probability-weighted assessments of brand trajectory from people who work with brands professionally.
"What probability would you assign to Brand X maintaining its current on-premise placement rate in New York through the end of the year?" Run that question with 75 participants who have relevant market knowledge, and you get a more honest answer than you'd get from any brand's own sales team.
Brands that are losing trade support often know it qualitatively before they see it quantitatively. A prediction market gives them a structured way to surface that signal earlier.
The use cases I'm building toward
Three specific applications I'm developing within Liquor Bets:
Portfolio allocation signals for buyers. A quarterly market on category performance expectations, designed to give retail and on-premise buyers a forward-looking sense of where category momentum is heading in the next 6 months. Not a recommendation. A calibrated probability from people who are watching the same market.
Brand equity tracking. Running consistent questions about specific brands over time to generate a longitudinal signal of how informed market participants view brand trajectory. The trend line over 12 markets is more valuable than any single data point.
Opening prediction for specific markets. Combining the BuildoutFeed pipeline data with crowd predictions about which openings in specific neighborhoods will succeed at which price tier. This is more experimental, but the underlying data is there.
What this doesn't replace
I want to be clear: prediction markets don't replace sales data, consumer research, or operational knowledge. They add a layer that those tools don't provide.
Sales data tells you what happened in your accounts. Consumer research tells you what consumers say they prefer. Operational knowledge tells you what you've seen work in practice. A prediction market tells you what informed people expect to happen next.
All four layers together give you a more complete picture than any one of them does alone. The hospitality industry currently runs on two of the four, and the missing two are the forward-looking ones.
That asymmetry is the opportunity.
