Opinion — South Australia Power Spot (NEM) (SA_DA) moved -52143.3% — Weekend Edition

South Australia spent part of 29 May paying people to take its electricity. The day-ahead spot didn't just dip — it printed a −52,143.3% move, the mathematical signature of a price that started the day positive and ended it deep below zero, somewhere near −$26,000/MWh against a morning reference around $50. That is not a typo and it is not, by SA standards, even rare. It is the most honest number the National Electricity Market produced all month, and almost everyone is about to read it wrong. Here is the misread. The number that will travel is the soft one: wholesale prices down 37.8% year-on-year, cheaper bills, the renewables-are-working narrative. That figure is true and it is boring. It tells you the average came down. The −52,143% tells you something the average actively hides — the variance went vertical. SA's price didn't fall so much as it tore itself in half. Midday collapsed toward worthless and then through the floor; the evening held firm. Those are two different markets wearing the same daily settlement, and the gap between them is the entire trade. Look at where SA actually closed on Friday: $119.86, up 9.8% on the day, one of the most expensive regions in the NEM. Victoria, next door, fell 48.3% to $51.67. So within the same grid, in the same week, South Australia was both the place giving power away at noon and the place charging the most into the evening ramp. The daily mean launders that. The intraday range is the signal. The forward curve already knows this, which is the part that should stop you. If SA's negative midday tail were the dominant story, you would expect the market to price its base load below calmer Victoria. It does the opposite. SA Base Q+1 trades at $82.00 against Victoria's $61.25; out at CY+1, SA holds $81.00 versus Victoria's $66.81. The market is looking straight at the region that hits −$1,000/MWh and beyond, and pricing its forward base a full twenty-plus dollars *richer* than its neighbour. It can only do that if the evening scarcity is worth more than the midday glut is worth nothing. Value isn't disappearing from South Australia. It's migrating — out of the middle of the day and into the 5-to-8pm window where the sun is gone and something dispatchable has to clear the market. That migration is the actual asset story. Rooftop solar in SA has grown roughly 15% year-on-year to more than 1.5 GW of unregistered, must-run, price-insensitive supply, sitting on top of a wind fleet near 2.5 GW that was running close to flat out on 29 May. Solar's capture price — what a marginal panel actually earns for the energy it produces — is collapsing toward zero, because every new array generates hardest at the exact hour the price is most negative. The money has left midday. It now lives entirely in whatever can ramp into the evening: gas peakers, hydro, and batteries discharging into the scarcity. You can see the shape of that premium one state over, where NSW peak forwards trade at $159.32 against $82.40 base — nearly double. The peak-to-base spread, not the average, is where the franchise is. Which is why the percentage itself matters beyond the headline. A −52,143% return is proof that every return-based model touching these nodes is now lying to its owner. Percentage change is undefined across zero; volatility estimates, momentum signals, and any VaR built on power returns either blow up or, worse, silently corrupt in precisely the negative-price regime they most need to measure. If your SA risk dashboard is still speaking in percent, it stopped working the moment it mattered. The fix is unglamorous and obvious: rebase everything to absolute $/MWh, and trade the peak-minus-offpeak spread directly, because that spread is the only thing on these nodes that stays well-defined when the level crosses zero. The reflexive answer is "build batteries — they'll arbitrage the tail away." I'd push back hard, because the evidence is already in. Battery discharge across the NEM more than doubled year-on-year, and South Australia still spent real time pinned near the −$1,000/MWh administered floor on 29 May. More storage did not stop it. The binding constraint that day wasn't how much energy SA could store — it was how much it could *move*, and the Heywood interconnector to Victoria was at capacity, leaving the surplus nowhere to go but down through the price floor. Storage depth doesn't help when the wire is full. And there's a trap inside the battery thesis that the build-out crowd keeps walking into. Every battery installed to monetise SA's negative midday tail charges in that tail — which lifts the midday price toward zero and compresses the very spread that justified the investment. The arb finances the fleet that kills the arb. As SA storage scales, the midday discount it feeds on shrinks, and the terminal value of a battery there stops being set by how cheap noon gets and starts being set by two things only: interconnector capacity and the size of the evening ramp. The forwards are quietly telling you that ramp isn't going anywhere soon — SA base barely slopes from $82.00 next quarter to $81.00 next calendar year. The market does not expect the evening premium to erode. It expects the cannibalisation to hit midday and leave the peak intact. So take the position the headline won't. South Australia is not getting cheaper in any way a trader should care about. It is getting *more bimodal* — average bills down, peak value up, and the entire economic surface of the day collapsing into a few hours after sunset. The −52,143% is not a glitch to be filtered out of the dataset; in the most solar-saturated grid on earth, it is the normal state, and it is the clearest available proof of where the money went. Price the spread, not the level. Own the ramp, not the average. And throw out any model still quoting these nodes in percent — on 29 May, it was already lying to you.