Why We're Building Elastics for the Thesis Trader
Source: http://x.com/i/article/2047640818748428288
Prediction markets allow you to turn your conviction and judgment into profit. At the same time, your trades benefit the public by providing your insight about the probability of a given event. The thesis trader, whose edge is judgment, deserve tools that extend and allow for a better judgment, while automating everything that does not contribute to it.
Prediction markets are the shortest path from a view about the world to a financial position on that view. No Greeks, no margin math, and no vol surfaces.
You just ask how likely is something to happen and about the price.
Meanwhile, every well-placed bet does two things at once: it compensates the person who was right, and it moves the market's price closer to the truth. Private edge and judgment become public information. This is why prediction markets matter beyond the people trading them.
Who is the Thesis Trader
The Thesis Trader is not the gambler, impulsive sports bettor, headline-chaser, crypto-degen, nor the quant running APIs. The thesis trader forms views on e.g. Iran, the Fed, an election, and a policy, and expresses them through contracts. Geopolitics, macro, politics, economy, crypto, weather, sports and other events.
Serious Thesis Traders on prediction markets allocate 50%+ or even more of their volume to thesis categories, and they are all growing.
Serious money follows good judgment.
The workflow of the Thesis Trader usually has six steps: monitor → form thesis → find contracts → size → monitor for invalidation → exit.
One step, that is, forming the thesis, is where the edge lives. It is the only irreducibly human step. Everything else is work around that step.
Judgment is the essence of what the Thesis Trader does
The Thesis Trader running dozens of positions can spend a lot of time every day on what isn't judgment: 15 browser tabs, manual price comparison across exchanges, checking whether a headline broke a thesis, recalculating exposure by hand, building their own spreadsheets or vibe-coding tools.
The edge, that is the actual cognitive work, is in the foreground or gets compressed into whatever time is left over. This is the core problem.
The scarcest resource the Thesis Trader has is attention directed at judgment. Everything else is friction.
What AI can do, and what it cannot
Beyond very short time horizons, AI is not a good forecaster. We don't believe it is, and we won't pretend it is. Good forecasting requires weighing what matters, down-weighting what doesn't, and integrating context that isn't in any training set. That's human work.
What AI is extraordinarily good is going wide. Monitoring thousands of sources simultaneously. Watching 100 positions at once. Executing a rule the same way at 3 am as at noon. Searching for stuff in huge datasets.
Terence Tao recently framed it perfectly: AI is good at going wide, humans at going deep. That division of labor is the future of Human-AI collaboration (or in words of
Ethan Mollick: co-intelligence
) and what we design around.
So the job is not to build an AI that trades for you. The job is to build one that extends your attention and absorbs the repetitive tasks, so the attention you have left lands on judgment.
Outside of pure quant trading, we don't believe in fully automated AI trading with serious money unless it has proper oversight, is controllable, and acts in line with human judgment.
When AI acts, it must be controllable, predictable, and accountable, with hard-coded and editable rules at various levels, including full logs and the ability to pause at any time.
No black box between you and your money.
What we're actually building
At
Elastics, we map the six-step workflow of the Thesis Trader and look at what we call the automation surface. This means the steps that can be expressed as rules without requiring judgment. Four of the six steps are automatable:
Monitoring,
finding contracts across exchanges,
watching for thesis-breaking signals,
managing exit.
Sizing could be partially decided or suggested by AI or entirely by a human. However, forming the thesis is entirely yours.
Our first agents and workflows are exactly this. Our AI searches for the contracts according to your criteria and rules.
You can assemble your own basket and portfolios of contracts to monitor and take action on.
For instance, a simple limit order, but scaled on multiple positions: IF the price drops below 25% for a pre-defined type of or selected contracts, buy $200. IF the position's PnL goes above 30%, take profit. IF it loses more than 20%, close it. IF the probability moves by X% in Y time, notify or exit.
Single-condition, deterministic, auditable. Nothing mystical.
These are the building blocks. Then added portfolio intelligence, alerts, news, correlations, portfolio exposure, whale movements, and all else you need.
Later iterations add multi-condition triggers, portfolio-level rules, and a signal layer on top, so the same architecture can act on news and surface contracts mispriced across exchanges. We start here because this is the layer that protects capital and captures moves while you sleep. Everything else builds on it.
This is all simple, deterministic, rule-based. Nothing mystical, opaque, or inherently risky and prone to hallucinations.
If you zoom out, what we're really building is an attention-extending system. This means a single environment where the Thesis Trader does not have to leave to do any part of the job. News monitoring, thesis formation support, cross-exchange contract discovery, resolution criteria analysis, sizing suggestion, execution, position monitoring, adjustment, and exit. All in one place.
We believe that every minute saved switching context is a minute returned to mental focus and judgment.
Why this is the template
Prediction markets are where this partnership is most obvious, because PMs are the cleanest translation of a view into a position. No intermediation and no derivative layer. No options model between your conviction and your P&L. For anyone with sharp views on the world but limited financial machinery, they are also the most accessible entry point into real markets. We take that seriously. Part of our job is to be an educator: to make it legible how these markets work, and how a view becomes a trade.
The cognitive workflow we're building for, monitor, judge, express, manage, is not unique to prediction markets. It's the workflow of every discretionary macro trader. Over time, the same architecture extends to TradFi. Same thinking, but different venues. But PM is where we start, because this is where a sharp view can still become a trade without a wall of infrastructure in between.
Closing
The future of serious trading is not an AI that replaces humans. It is a human whose attention is amplified by AI. We are building for the person whose edge is judgment, and whose scarcest resource is the attention to apply it. If that's you, we are building this for you. Check at