At the moment of truth in a negotiation - the other side names a number, and the room goes quiet - nobody in it is rational. Not me, and not you. I have run hundreds of negotiations over my career, with suppliers, customers, partners, employees, banks and lawyers, and our hands shake in exactly the places where two Nobel laureates told us in advance they would shake. The interesting question in 2026 is what happens when you put a third party in the room that reads both of them and does not shake.
What Kahneman would say about your last negotiation
Daniel Kahneman, winner of the 2002 Nobel Prize in Economics, spent a career mapping exactly where human judgment breaks. Lay his findings over the last deal you closed.
Anchoring. The first number on the table drags the whole discussion toward it, even when it is completely arbitrary. Whoever named a number first drew the field, and you negotiated everything after that relative to it without noticing.
Loss aversion. A loss hurts roughly twice as much as an equivalent gain pleases. That is why, at the end, you gave up a clause worth real money rather than "lose the deal" - when on a cold calculation, walking home was the better outcome.
The endowment effect. The moment you started picturing the deal as done, it became yours, and what you were willing to concede to keep it jumped. If the other side is experienced, they read that on your face.
The problem is not that we do not know about the biases. Kahneman said that after fifty years of research he still fell into them himself. The problem is that you cannot see them from the inside, in real time, with the adrenaline running.
What Nash adds to the equation
John Nash, Nobel 1994, looked at the other face of the same coin: not how we go wrong, but what the confrontation actually is as a structure. Game theory gives a negotiation a cold skeleton - each side's real walk-away point (the BATNA, the best alternative to a negotiated agreement), where the zone of possible agreement sits, what each side genuinely loses if the talks collapse, and which moves are a credible threat rather than a bluff.
Most executives I have met negotiate on feel. They have never written down, in numbers, what their own alternative is and what the other side's alternative probably is. That is the information that decides the outcome, far more than charisma, experience or chemistry. On a deal worth tens of millions, the gap between that preparation and a gut feeling can be worth more than every consulting fee you have paid in your life.
Putting the two together is the whole point. Nash tells you where you ought to be standing. Kahneman explains why you are not standing there. A system that holds both at once is worth more than the sum of its parts.
The whole theory in one 10-shekel bottle of Coke
It is hot. You are on the beach, dying for a drink, and the kiosk sells a bottle of Coke for 10 shekels.
The real arithmetic, the way Nash would run it: what is that Coke worth to you right now? Say you would pay up to 15 shekels before bothering to walk fifteen minutes in the sun to the supermarket. The kiosk owner paid 5 for the bottle. So every price between 5 and 15 is good for both of you: at 10 he makes money, and you get a cold drink worth more to you than what you paid, without the walk. Both sides come out ahead. That is Nash on one foot - there is a range where closing beats not closing, and the deal is supposed to land inside it.
And what happens in practice, the way Kahneman would predict: yesterday you bought the same bottle at the supermarket for 7. So when the kiosk says 10, you do not think "worth 15 to me, excellent." You think "yesterday this cost 7 - he is ripping me off by 3." And whatever feels like a loss stings double, so those three shekels feel like six. You say "nobody rips me off," and you walk fifteen minutes in the sun, thirsty, to save 3 shekels.
Look at what just happened: you gave up a good deal - a Coke worth 15 to you, for 10 - and you paid for it with fifteen minutes of sun and thirst, just to avoid feeling like a sucker over 3 shekels that exist only inside your head. The supermarket price from yesterday is irrelevant to anything: you are not at the supermarket, you are on the beach. The decision was not driven by what the thing is actually worth to you. It was driven by the number in your head.
Now the trick. Imagine the kiosk had a sign: "At the hotel across the road - Coke, 25." Suddenly 10 is a bargain. Nothing real changed: same bottle, same price, same supermarket fifteen minutes away. Only the number in your head got swapped, from 7 to 25 - and it cost the kiosk owner nothing. Richard Thaler, the third Nobel laureate of this story (2017), showed exactly this in a famous experiment: people on a beach agreed to pay nearly double for the very same beer once told it came from a fancy resort hotel rather than a run-down grocery store.
Now swap the bottle for a deal worth a million. It is precisely the same mechanism, except the thirst is called "cash flow," the hotel sign is called "an opening offer," and the fifteen minutes in the sun is called "blowing up the negotiation." Which raises the obvious question:
So can they be joined in a single formula?
They can. And what is surprising is how little has to change. Nash looks for the split that maximizes the product of both sides' gains above their walk-away points:
The model quietly makes one assumption: that each side's psychological zero is its BATNA. And that is precisely the assumption Kahneman demolished. People carry a reference point - the first offer they heard, last year's contract, the number they told themselves on the way in - and it is almost never their true walk-away point.
The fix is one line (in the math block at the end of the section): keep Nash's product, but measure every gain through Kahneman and Tversky's value function, from the reference point r instead of from the BATNA:
Everything Kahneman discovered enters through the measurement. When the reference point sits above the BATNA, the whole region between them is experienced as loss - and multiplied by loss aversion, times 2.25. A side that anchored itself high feels it is giving up more than twice what it is actually giving up. That explains the most common failure in negotiation: deals that die inside a genuine zone of agreement.
And the combined formula predicts three things Nash alone cannot produce. First, the reference point becomes a strategic variable: whoever sets it moves the split without changing anyone's actual alternatives - that is the opening offer, as a number rather than a feeling. Second, a side sitting below its own reference point turns risk-seeking, preferring to gamble on blowing up the deal over accepting a certain loss - the executive who walks away from a good deal. Third, the less loss-averse side captures more of the pie. The combination has an academic literature - work on Nash bargaining with loss aversion - but the operational conclusion needs no journal:
Compare the sum of the two walk-away points to the size of the pie, and the sum of the two reference points to the size of the pie. If the walk-aways fit and the references do not - there is a real deal on the table and yet both sides feel short. The gap is anchoring, not money. Hence the practical rule: moving a reference point costs nothing. Buying someone past their loss aversion costs 2.25 to one. Before you pay, check whether you can simply move the reference.
You can also see all of it in one picture. Take a pie of 100 and two sides whose walk-away points are both zero. When both measure from their walk-away points, Nash's product peaks exactly in the middle - fifty-fifty. Now let one side anchor itself at "not a shekel below 30": walking away now hurts it 2.25 times as much, the formula prices in the desperation - and the peak moves against it, from 50 to 31. Anchoring yourself above your walk-away point is weakness, not strength. You throw anchors to move the other side's reference point; your own stays glued to your real BATNA.
And honestly, because this whole article stands on honesty: 2.25 is a laboratory average, not a property of the person sitting across from you. The formula is a lens that produces the right questions and sensitivity scenarios - not an oracle that spits out a price. Anyone selling it as an oracle is selling you a story.
Want the full mathematical breakdown? The formulas are here
The Nash bargaining product - the split that maximizes the product of gains above each side's disagreement point
The Kahneman-Tversky value function - every gain is measured through it, from the reference point
The combined formula - the Nash product, measured through Kahneman's value function
What a negotiation support system looks like
Picture it. Before the meeting you hand the system the picture: the offers exchanged so far, your own constraints, whatever is known about the other side. It builds a map of the situation - the estimated zone of agreement, the value of each scenario in money, your walk-away point written down in black and white, so that under pressure you do not invent a worse one.
During the negotiation, after every round, it updates: their offer moved by 4 percent, your concession was double theirs, you are reaching your red line faster than you planned. And the part that matters most, it catches your biases and names them out loud. "You are anchored to the number they threw out in the first meeting." "That concession looks like loss aversion, not like arithmetic." A blunt friend with a doctorate and no ego in the room.
At the end, a simulation: three scenarios for where this goes, each with a probability and a price attached, before you send the next email.
Be clear about what the system is not. It does not replace the executive in the room, it does not send email on your behalf, and it does not run the negotiation. Every decision stays yours. It does the one thing Kahneman recommended that nobody ever finds the time to actually do: slow the thinking down at the moments when speed kills, and attach to every gut feeling a number you can argue with. Kahneman called it the outside view - seeing your case as one of a hundred similar cases instead of a uniquely special story. People are consistently bad at this. A machine is excellent at it.
Why now, and why I am writing about it
Until recently this was science fiction, because the knowledge sat in books and in the heads of expensive advisors, and reaching it at the moment of truth was close to impossible. Language models changed the picture: they can read an entire negotiation correspondence, pull the positions and the moves out of it, and cross it with decades of decision-making research - in seconds, and in Hebrew.
I will say this carefully. For me it is no longer theory. There is a working model, it has sat through a real negotiation, and what I saw convinced me this is one of the most interesting categories that will be built in the coming years. The details stay, for now, in a closed room.
If you have a significant deal on the table - a merger, a large contract, a substantial lease, a dispute heading toward mediation - and you are curious how this looks, send me a message. I would be glad to show you a demo, in full discretion.
