Model Update – Cal Hydro

Good Afternoon,

Ansergy implemented two changes over the weekend:

  1. Added more Hydro Stations to California – this change provides more resolution to the hydro energy forecast for California and has resulted in price changes to the forecast.
  2. Smoothed the Demand Forecast – our model breaks a year into 24 temperature periods with each using a unique set of algorithms to solve for demand.  The transition from one period to another sometimes resulted in noticeable demand steps; this change removes those steps.  There was a very small change in price for all hubs in our term forecast.

We would like to believe our forecast is perfect, that our models could never be improved, but we know that not to be the case.  No model is perfect and every model can always be bettered.  Our commitment to our clients is to continually improve upon our forecasts, and sometimes that requires a change in algorithms, new data sources, or new models.

California Hydro Stations

The old model ran two stations, one in SP and the other in NP.  The new model is using six at NP and four at SP.

NP Hydro Forecast (On Peak):

000_NPhydro

The new forecast is more stable and, in most periods, higher.  The spikes in the old forecast were a result of using  a single station.

NP Price Impact (With Change vs Without Change):

000_PRICEnp

Price impact is obviously bearish where there is more energy and bullish where less.  Less surprising is the relative minor impact of this change on prices, except for summer, which is materially higher.

SP Hydro Forecast (On Peak):

000_SPhydro

The new forecast is lower across the board which is due to using more hydro groups which use different stream flow proxies.  The wild card at SP is the Big Creek complex which has no real-time or historical sensors.  The one which was originally deployed was over-stating SP hydro energy.

SP Price Impact

000_PRICEsp

Change in Demand Forecast – Mid-C

000_MCdem

The front (days 0 to 15) was not affected by the model change, the volatility there is explained by changing weather forecasts.  The change in the back of the forecast is minor, almost imperceptible, though it does deliver a much smoother curve.

WECC Demand Change

000_WECCdem

The change in aggregate (WECC) is more apparent, especially in the transition from June to July.

Drop me a note if you have any questions or comments ([email protected]).

Mike