GTO Turn Charts: How Is Chart Built?
The turn card is where most solvers bleed EV, and the chart construction process is the reason. Building a GTO turn chart is not a spreadsheet exercise. It is a brutal filtering operation where you take the flop strategy tree, lock the turn card, and re-solve every combo against a polarized range that no human can memorize. 🎯
The Node Locking Trap
Every turn chart starts with node locking. You fix the flop action frequencies, then force the solver to treat the turn as a fresh decision point. The problem: if your flop locking is off by 2%, the turn EVs cascade into garbage. I have audited databases where the turn barrel frequency was inflated by 14% because the flop check-raise range was missing 9 combos. That is not a rounding error. That is a chart construction failure that costs real money at 200NL and above. 💀
The fix is ugly but mandatory: re-run the flop solve with tighter bet sizing buckets, export the turn decision nodes, and rebuild the chart from scratch. No shortcuts. No copy-paste from preflop ranges.
Equity Distribution and Blockers
Turn charts are built on equity distribution, not raw hand strength. The solver calculates how each combo performs against the opponent's entire continuing range after the turn card lands. Blockers matter more here than on any other street. A hand that blocks the opponent's nut flush draws gains massive EV on a blank turn, even if its absolute strength is marginal.
You must isolate blocker effects into separate chart columns. If your turn chart treats a hand with the ace of hearts the same as the same rank without it, the chart is dead on arrival. The difference in barrel frequency between those two combos can exceed 40% on a wet turn.
Bet Sizing Buckets and Frequency Mapping
GTO turn charts use discrete bet sizing buckets. The solver picks from a fixed set: 33%, 75%, 125% pot, and all-in. The chart then maps frequency percentages to each bucket. This is where the algorithm gets violent. A hand that bets 33% pot 60% of the time and checks 40% is not the same as a hand that bets 75% pot 30% and checks 70%. The EV difference between those two strategies on the turn is often less than 0.3bb, but the variance difference is enormous.
Smart chart builders collapse low-frequency, low-EV actions into the nearest high-frequency bucket. This reduces chart complexity without sacrificing EV. If you are still using six sizing options on the turn, you are overfitting.
Range vs Range Interaction on the Turn
The turn card changes the nut advantage. A card that completes the flush gives the preflop caller a massive range advantage, while a blank card keeps the preflop aggressor in control. Your turn chart must reflect this shift in real time. The solver does it automatically. Human chart builders often miss it because they anchor to flop frequencies.
Audit your turn chart by checking the check-raise frequency on flush-completing turns. If it is below 12% for the caller, your chart is too passive. If it is above 25%, you are over-bluffing and burning money against calling stations.
Tree Size and Computational Limits
A full GTO turn chart for a single flop texture requires solving 1,326 combos across 49 possible turn cards. That is 64,974 decision nodes. Each node needs equity calculation against a weighted range. The computational load is why most commercial solvers truncate the tree at the turn. They give you a simplified chart and hope you do not notice the missing branches.
Real turn charts are built with cloud clusters running 256GB RAM minimum. If your chart was generated on a laptop, it is a toy. The difference between a toy chart and a production chart is the difference between breaking even and crushing 500NL.
Final Calibration
After the solver spits out the turn chart, you must calibrate it against population tendencies. Pure GTO frequencies assume a perfect opponent. Real opponents over-fold turns by 8-12% in low-stakes games. Adjust your barrel frequency upward by that margin. Against high-stakes regs, drop it back to baseline.
The chart is never finished. Every new solver update, every meta shift, every rake change forces a rebuild. If you are not rebuilding your turn charts quarterly, you are playing a solved game with last year's solution. The edge is gone.