XENEURO · a spiking frog brain hunting new pairs

Plate I · anatomy

The same frog, cut open

Twelve sections from snout to toes, lit by the same spikes as the frog above. Drag the cut along the body; click a neuron to see what it hears and what it sends.

fig. 2 · axial · z = ——

Rdorsal ↑ · live activity in the slice · click a neuronL

W 1.00 · L 0.50drag image: contrast ↔ · brightness ↕

0.00cm

—

series · 12 sections, snout → toesclick to jump

fig. 2b · sagittal · along the midline, neurons projected · drag to move the cut

equity–

P&L–

hit rate–

swallowed–

digested–

spat out–

flies seen–

alive for–

Abstract

A frog that hunts by feeling the water

Xenopus laevis has no tongue. It lives in murky water and finds insects that fall on the surface with its lateral line — rows of stitch-like sensors along its back that feel ripples and tell it where they came from. It lunges, scoops with its clawed hands and sucks the prey in.

Here every fresh trading pair on the chain lands on the pond as a fly, and every trade in it makes the fly buzz. The spiking neurons decide what to scoop, what to spit out, and learn from what happens next. Real market data. Nobody is driving.

Plate II · live

The pond

A fly is a fresh pair. Its ripples are trades. Green smells good to the frog, red smells wrong. Hover a fly to see what the frog feels; click to open its chart.

Fig. 3 · dorsal survey of the pond. Frogs are outlined in their own colour, the one on the table is bright; rings are trades; the dashed line joins it to the fly it hunts.

Plate III · selection

The frogs

Same brain anatomy, different genes. They hunt the same flies; the worst is retired by its meals, not by the clock. Click a frog to put it on the table above.

Plate IV · inside

The nervous system

Every row is one neuron, every tick a spike, the last three seconds. Senses on top, the brain in the middle, muscles at the bottom.

spikes / s–

swim beat–Hz

dopamine–Hz

feeding–Hz

nausea–Hz

satiety–Hz

weight drift–%

flies on water–

Fig. 4 · spike raster, all neurons, sliding 3 s window.

Fig. 5

Taste · what it has learned

Learned ÷ anatomical strength of each smell cell's synapse onto the feeding command. Nobody sets these bars: dopamine does.

Fig. 6

Stomach ·

Open positions. The gut reports their profit and loss back to the brain through vagal afferents.

empty stomach

Plate V ·

Track record

Every swallowed fly, bought and sold. All results are net of pool fees, gas, price impact and token taxes on both legs; buys fill at the next quote. Dollar amounts are shown on a $1,000 bank: the live frogs trade $100 and size each buy as a share of equity, so every dollar here is ×10 and every percentage is exact (fixed costs such as gas are scaled too; on a real $1,000 bank they would weigh less). export CSV ↓ (raw, $100 bank)

Fig. 7

Equity

Fig. 8

Meal by meal

bar = net result of one trade · dashed outline = before fees, gas and impact · band = streak

#	closed	fly	bought	mcap at entry	how	held	buy → sell	gross	pool fees	gas	impact & tax	net	streak	running

Erratum · the DexScreener era

The profit that wasn't

Before our own chain indexer, the frogs were priced with DexScreener. These are the best results we ever had on screen, kept exactly as they were shown, next to the same trades re-priced against the chain's own swaps.

Correction. The numbers marked “on screen” were real outputs of our software, but they were measured on prices that were about 17 s old. They overstated the profit. Everything else on this page is priced from the chain.

on screen · best frog · DexScreener prices––

realised, as recorded––

realised, re-priced on chain––

Fig. 9

What the screen showed

equity of the best frog, every 30 s, open positions valued at DexScreener's last price

Fig. 10

The whole pond

Every frog of that pond: what its bank showed at the end, and its closed trades as recorded and on chain.

frog	bank on screen	on screen	realised · recorded	realised · on chain	checked

Fig. 11

Closed trades, one by one ·

as recorded re-priced on chain could not be checked (kept as recorded)

Fig. 12

Where the gap came from

Average price move from fill to exit, same trades, same timestamps, all frogs of the run.

Fig. 13

Why, and what changed

Prices arrived late. DexScreener lagged the chain by 17 s (median). Exits fired on stale peaks and dips, so digests looked richer and spits looked worse than the prices anyone could actually get.
Open positions were marked at stale prices. The bank on screen counted them at DexScreener's last price, including pools that were already emptied.
New pools appeared late or never. Median 63 s after creation, 4 of 12 never listed.
Now: our own chain indexer (new pools in 5 s, 5.6× more price updates), buys fill at the next quote, sells at the end-of-block state. Calibrated on the next run: recorded minus chain price move, median 0.0%.

Plate VI · senses

The water

Live pairs from DexScreener and GeckoTerminal, contract checks from GoPlus, the developer's history read from the chain.

fly	age	liquidity	mcap	trades 5m	5m	1h	developer	red flags	smell

Plate VII · narrated

Activity log

Methods

A real circuit, honestly labelled

Fly brains have connectomes traced from electron micrographs. Nobody has traced a Xenopus brain that way yet, so this frog is built from what is known about it: which cell types exist, what they connect to, and how the animal actually hunts. Every spike is simulated, every trade uses real market data, and every rule the rig imposes is logged as SYSTEM.

M1The circuit

Swim network: dIN, cIN, aIN, dlc, Rohon-Beard and motor neurons, wired by the rules measured in hatchling tadpoles (Roberts, Soffe, Borisyuk et al.). The rhythm comes from commissural inhibition and rebound; no oscillator is coded in. Prey localisation: lateral line → medial octavolateral nucleus → torus semicircularis → reticulospinal turn (Elepfandt; Claas & Münz).

M2The neurons

Leaky integrate-and-fire neurons simulated at 1 ms resolution in real time.

Fast AMPA-like, slow NMDA-like and inhibitory synapses; spike-frequency adaptation, bout fatigue and rebound currents. Learning is dopamine-gated with synaptic tagging: tagged at the gulp, reinforced when the outcome arrives minutes later.

M3The body you see

The frog in fig. 1 and the sections in fig. 2 are a schematic model built from simple shapes, not a scan; the organs in the sections (skull, vertebrae, heart, liver, gut, kidneys, fat bodies, lymph sacs) are drawn from textbook anatomy with MRI-like contrast, not imaged. The neurons are placed where their real counterparts sit — bulbs at the snout, torus in the midbrain, swim network down the cord, gut in the belly — and each lights up when that simulated neuron spikes. Flies above the frog are real pairs, placed by their direction from it; distance is compressed.

M4Limitations

It is not a traced Xenopus connectome, and we never claim one. No real funds are traded: every trade is simulated at live prices, with price impact and fees. It is not financial advice. The frog is a frog.

inSensestrades per 5 min → ripple → lateral line
buy pressure, momentum, freshness → appetitive smell
selling, dumping, thin liquidity → aversive smell
position P&L → vagal afferents

→BrainMON → torus map → reticulospinal left/right
smell → feeding command (plastic)
gut → satiety, area postrema
dopamine gates learning

outMusclesswim network → motor neurons → thrust and steering
suction + limb burst with prey in reach → swallow = buy
spit burst → spit = sell
satiety → digest = take profit

On-chain

$XENEURO

Launching on the chain. The contract address will appear here and nowhere else.

CA · coming soon