Benelux Intraday Price and Balancing Forecasts | Vincent Thevenin, Montel

Good afternoon again, really happy to be here. So my name is Vincent Tevenin, market expert and mentor. Today is a topic of my presentation, will be about shorter markets and without waiting. Let's have a look at the content of my presentation. There will be two parts. So on the left, there will be a part about the intraday price index forecast, focus about Netherlands. And on the right, I will talk on the right, I will talk on the second part about the balancing forecast, the rise of passive balancing and focus on Belgium. Without waiting, maybe you were asking yourself, what is the intraday price index? Why do we need that? So the why do we need that is that intraday, I think if you're sitting in this room, you know it's much more complex than day ahead. You do not have one price at the same, it's not like the head auction on Hydera, it's more complex than that. So you do have many trades happening at different points in time with different prices. So a way to summarise this whole complexity is to define some indexes. An easy index is just to take a volume weighted average among a given time. And here on screen, you have different ways. Why do we need this index? So the first one would be for valuation. If you are building a new asset, we have just done before, you want to know what are the intraday prices. I can expect a second use case of these indexes as benchmarking. So I used to do intraday shift myself. So the way to define the P&R, if I were good or not, were to use intraday indexes for me as a benchmark. So I am a good trader or not. And then finally, two further use cases that are tackled at the same time. So it can also be used for financial settlement of hedging. So if you are handing over a position from the day head to the intraday desk, you can use, for example, the indexes as a way to passing over the position. And we have heard there's still no futures yet about intraday indexes. Maybe we can't, but it could be useful for hedging as where indexes. In terms of indexes, you have different kinds that are existing today. You have the ID.1, ID.3, ID.4. And the main differentiation between these indexes is how long, what is the concert time period behind. ID.1 uses the last 60 minutes of trading, both in Belgium and Netherlands, a bit different in Germany. ID.2 is the last three hours, ID.4, the full session. And depending on what is your portfolio, if you're a spec trader or if you have rests in your portfolio, then the indexes would have a different weight for you. So ID.1, I would say, now is the most important one because most of the trading is happening at the very end, where you have liquidity and most important indexes. Let's meet now on screen our latest products, the intraday price forecast in Enapsis. Today, I'm going to focus about Belgium and Netherlands, also available in many different countries. So on screen, you can see intraday price index for quarterly hourly products, but it also exists for hourly products and with a drop-ton QLVA voltage switch between the indexes. So what are we proposing? We are creating intraday price forecast starting from two hours before delivery, also one hour before delivery and up to 50 minutes before delivery. We also are calculating some sensitivities around that, so P50, P80, P95. And what's also important and also to measure against is the outturn, what was the index. The indexes are usually published by the poor exchanges on D plus one, but we are also calculating during the day on D with the on-training data we have. And now what I'm trying to do with you, so I said I used to be a trader, so I'm going to try to take you by the end. And how for you, how to make money with our data, I think if you are a trader, that's what interests you to make some P&L. And so on the next slides, I'm going to explain you how to take our data. So from the forecast we are creating, how to create trading strategies out of it, and how to assess if there are good ones or not. So here for all what I've represented, I will use a one megawatt trade as an exemption per co-actor. I will also, it will be net of exchange fees. And when you want to trade, create a trading strategies, you have different steps you need to follow. It depends if you have an asset or if you are a spec trader, but you need to define first a signal. I will deep dive into that on the next slide. But just briefly speaking, you get, you create the signal out of the price forecast. We look at the gap between the price forecast we're creating with the concrete market prices and then out of it, you are derivating a position you want to have. So either to buy, to sell or do nothing. Then what you also need to assess is what, at which price can you enter this position at all? And what product we use? We use Quartal Ali or Ali products. And for that, we are using the snapshots of last traded prices from the market at the time of forecast creation. It's an important point. And in the end, if you want to exit this position, if you are doing in and out, you need an intraday price to measure that. And that leads to the formula in the middle. So the net PNL per trade is basically the direction multiplied by the exit minus the entry. So the net PNL per megawatt hour at the moment. Now I'm going to focus on the signal creation on the top left. So with the data we are producing, you have different ways to create different strategies. But for today, I'm going to focus on only one, which is taking the intraday price forecast for the quarter hourly granularity and taking a decision per quarter. So to explain how to create a signal. So on the left, you can see different quarters. In purple, this is the intraday price forecast that we are generating. For Q1, we have a price forecast of 110. We look at the markets at this time and the market was at 105. So our forecast is higher than the market. This is a buy suggestion. This is a signal for this quarter. You can generate this, of course, for the other quarters. And for Q2, you would sell. Q3 do nothing. Q4 sell again. So these are the signal creation you can generate from the forecast. Then you need to backtest it against real entry price, real market prices. And finally, to get the exit. Here to get the exit, I will further explain later on. You have different possibilities. Shall you take the index itself, which are actually more realistic, but you can also take the price at 30 minutes before delivery, 15 or the latest closest price. So without waiting, look at the results. So I'll not go to all numbers, but I will focus on the numbers which are in circles or ovals. So here is the P&L results backtested for the strategy only for quarters using our forecast. It's only a short, volatile three months for both Belgium and Netherlands. And each time, the assumption is to use a position of one megawatt per quarter. And what it shows, it does this in the middle, the result outturned as a spread. So it's a forecast as some value, some skills. It can allow you to get some spreads. And then again, the discussion, which exit price shall you take as a benchmark? I think the ID 1, the quarter is the fair one. And if you are translating this spread of the time period you were having for the three months, it would have been 14k for Belgium and around 10k for Netherlands using the price forecast. If you now take the same strategy, but you take a longer period only for Netherlands, here the backtesting data I've used is around 11 months. You can calculate this P&L and plot it here on screen. And the very same strategy for Netherlands would have brought you 76k if you used ID 1 price index as a benchmark. If you use different prices, like the closing price, it could even be higher than that, but might not be realistic. Maybe something to bear in mind. So most of the big chunk of the P&L, around 36%, is done on days, on quarters, where we have a big gap. So on high, there was really high prices. So this is something to bear in mind. Now I've shown some exciting numbers, but there are some caveats. I will be transparent about it. So the first one is liquidity, of course, because is there the liquidity to get into the position? I'm here to enter the position I've been showing you. So here on the left part of the slides, I've plotted you the liquidity in intraday in the last hour for different countries, for Germany, Netherlands and Belgium. And I think it must be known to you, but Germany is obviously the most liquid market. Netherlands is also somehow liquid, a fifth of Germany. But I think it's still realistic to enter this position here in and out. But Belgium is a bit here the problem because it's not liquid at all. That's the thing here. So what we're showing here, so you have two bars. The green one for the hourly and the gray one for the quarter hourly. So that's a big caveat for that. And one of the reasons why the numbers look like this is in Belgium, you can see on the right part of the slide, you can see the cross on a share of the volume. So as this volume, internal volume, so as this volume made, we have one leg, which is cross-border. And you do see here quite a big differentiation between the country. So Germany has only 24% of cross-border volumes, Netherlands 40%, 41%. And Belgium, most of the volume is done cross-border. So basically, there is a rule of thumb. So Belgium is quite liquid when it's connected to X-bit. When it's not, it's not liquid. So you might know that one way to reduce analysis again, I've not done it, would be to select the positions where the liquidity is available thanks to the X-bit capacities. And maybe one good point to end. So the Belgium liquidity might appear. We'll see why. So this year on January 14, there was a big market design change, namely the implementation of, I need to get it right, it's quite long. The 30 minutes cross-zone or gate closure time. And what does it mean? It means that today, so far, before the 14th of January, the lead time before delivery to trade cross was 60 minutes. It has been shortened to 30 minutes before delivery. And there are many. So on the map on the right, you can see that all the green dots are the borders which are concerned or remade the switch. There are still some borders coming in, including the borders for Netherlands and Germany. So on the table on the left, you can see that at the moment it might be delayed. But for Q4 2026, we can expect the German-Dutch border to switch as well. And for Q1, we can expect the borders with Q1 next year with Belgium to switch as well, not for France. France will come later on, likely in Q1 2029. But for all the switch, I think you can expect quite an increase of liquidity. I had a similar presentation in Austria in VIN a few weeks ago, and I have analyzed this. And for Austria, so since the go-live in the last hour, we've been witnessing 100% liquidity in the last hour. So this is really a big, important step. And so we can be optimistic that the Belgium liquidity will increase then. That was the first part of my presentation. I will now move to a second topic, this time focused about balancing in Belgium. And it's called the rise of passive balancing. Go on. Yeah. I'm not sure. Yeah, maybe this one. Yeah. I was wondering about the liquidity. Do you see liquidity? How do you see the liquidity resulting in P&L for your asset? Is more liquidity positive or negative? So here for the Belgium one, it will be positive because I don't think the numbers I've showed are executable. Right. I would say positive. Yeah, this is rather for a spec trading, so you can enter and exit more easily. Well, yeah. Any of you? Okay. We can discuss it in more detail. But thank you for the question. Hey, I'm Tore Mairena from TTP. So I was wondering if you also provide any kind of confidence interval for that intraday one forecast? Yes. Sorry, I wouldn't quickly sue that, but we do have, yeah, P98, P80, P50. So we have sensitivities. And it's not sure you can see, but it's a little bit in yellowish. So you have an exact deterministic value and a range around it. So, yeah. And how do the backtest show? I didn't do the backtest with that. But it's another dimension you can have on top of that. Okay. Thank you. Thank you for the question. And now moving on to the second part. Before deep diving into Belgium, I want to explain you a bit, explain the trend in Europe. So the rise of passive balancing. We're in Amsterdam. We're in Netherlands. So I want to explain you what passive balancing is. And it has been there for quite a while. Really happy about it. So it has been there. What has been happening this year is in Switzerland. So since January, we have a thing of pricing. It's the opening of passive balancing. So it's great, really happy about it. And the consumers as well. Because the balancing costs have been reduced by 40%. You have many BIPs entering this market. So quite interesting development here in Switzerland. And on the right part of this slide, you can see what's happening now or for the future. What is in the planning between next year? First one is Belgium. That would be the topic of my presentation. So we're not too much deep dive now. But there is a consultation ongoing on this topic. So if you have staked in it, I would recommend you to reply. There is also a final consultation, which will happen in Q2, 2027. And maybe a go live in Q4. So that's what I'm talking about next year. But we'll touch that in the next slides. And here in Germany, it's a big elephant in a room. So here are breaking news about that. Because actually yesterday in a German balancing service providers conference, it was announced for the first time by the TSO officially. So they want to enter strategically into the passive balancing and open it to relax. So it's quite a really good news. And here there is on a timeline, it will be the BRP contract need to be adapted. A lot of details to be sorted out. But the BRP contract shall be consulted in the second part of 2027. So generally speaking, we see this as a trend. And the common drivers for all these countries, we have more renewable energy sources being installed. And one way of tackling it is to let the market help. Because you can reduce the cost at the end. And you can contract less control power. So that is the idea behind that. Now, so here you go. Yeah, for sure. I think we really find it interesting. Because in the Netherlands, we see the passive balance actually causing a lot of problems in the ansal escalation. So that is pushing back all the problems. That is a good point. So Netherlands now maybe did too much. So you have a regulation state too, which is a problem. You have the A-soscillation. So, yeah, but it's still, so obviously it doesn't harm the other TSOs to go in. Yeah, it will be interesting. So the A-soscillation, it's quite an interesting topic. Because the Netherlands is as proposed. If it doesn't solve the problem to reduce the transparency. So I don't have a full head for you. But the discussion is not over for sure. Well, okay, moving on. So the rest of my presentation will be about Belgium. And I have two slides about explaining what are the expected regular changes. And then I will present here our product. So what are the plans from Elia? So Elia has done like a consultation. There's actually a big PDF explaining that. Just two slides, three rough summary. And there's an article, the 16.2 in the BRP contract. I can't maybe read that. It allows actually the BRPs to take some passive balancing position. But provided they keep the resources, they have to flex direct to come back if necessary. And so this second part actually has not really been enforced at all. So it has only been enforced for the pure traders. So the Elia colleagues are monitoring that. For a pure trader, you cannot take an imbalanced position. But if you have assets, it was more or less okay so far. So what they want to implement is to change the article and to have a new definition, which is so-called maximum authorized real-time exceedance. And so it leads the way to two different definitions. So what is a real-time exceedance? It's the part of the BRP imbalance that you can take if you are able to restore it in real-time. So if you have assets like battery of concentration flexibility, you can use it. But you need to be able to restore it if something's happened. And on the other part, which is this time valid for all BRPs, even for pure traders or without assets, there's a new definition, the maximize authorized real-time exceedance. And now what Elia is planning is to offer for all the BRPs up to 10 megawatts of passive balancing position. And this is quite a gate changer. And we will see here what's also some consequences. So it's also a risk. So it's a relaxation and opening. But Elia is planning also to have some monitoring to fight against the drawbacks. On the top left here of this draw, so what Elia is going to do, it's going to do a monthly monitoring of the position which are taken into passive balancing. And there is what is marked in yellow as there is some kind of a threshold that BRPs are doing passive balancing needs to meet. And when the system imbalance is bigger than 100 megawatts, the accuracy of this position needs to be higher than 60%. That's, let's say, the quality that BRPs need to achieve. And if it doesn't happen, if it's rich on a monthly basis, the BRP will receive a warning. And if a second warning is happening within 12 months, then BRP will not be able to use passive balancing anymore for the next 12 months. So it's a bit like a kill switch. The exceedance will be put to zero for the next 12 months. And on top of this monitoring, there are also some changes. So the collaterals can increase. So to reflect that, you need to look at the special rules. And also for the new BRPs, they will need to wait three months to have some track record on the performance before they can enter this game. And now moving on to our products, the balancing forecast. So what can you do about it? What are we offering to help you catch these regulatory changes? So on screen, you can see the NRV forecast in an episode. It's actually available for many countries and not only Belgium. And here in the middle, you have the different graphs, the curves we are producing. So on end, I will guide you through the boxes one, two, three, four. So the first output, the main output we're trying to forecast is the NRV, which is obviously one of the main drivers of the imbalance price. We have the forecast in blue and as usual in anapsis, the actual is in black. And on the top, what we're also forecasting and sometimes the imbalance prices, let's say the collateral output, with the forecast in blue and actual is in black and the entry prices in purple. And then what's also interesting on the right of the slide is when this forecast is created. So it's starting four hours before delivery, three hours, two hours, one hour, and also every 15 minutes before delivery. And there's also the risk tolerance parameters. We have the way to create a signal that will deep dive the next slides. We have a high risk, average risk and a low risk where, let's say, less position will be taken. You have a more secure approach, but we'll explain you in more detail on the next slide. Here on this slide, let's explain how do we do it and from the drivers to the signals. So we have you go from left to the right on the slide. First, the features we are seeing, generally speaking, we are quite transparent in how to do it. So we use wind, PV, demand, direct, interest price, NRV. And at the bottom, you can see the drivers, megawatts on the top, the price, the time series we are using. And then from these features in the model, machine learning model, we are computing a forecast probability direction. And it was highlighted in the middle by different colors. So the probability of the system to be short, to be neutral. This is, by the way, neutral where this risk is happening. Risk parameter and percentile for system long. And out of this probability, that's again similar to the intraday signal duration, it creates a recommendation. So either to buy, to sip, do nothing or to sell. And this is can, yeah, if it's a forecasting system short, it will recommend you to buy. That's where it works. And here, in the same way as the previous product, I think the best way to talk to traders is to show them some P&L. So that's why, as a way I used to do a few work. So I'm doing the same now. So following the signal, I've been producing a positive P&L for a long time period. I took here for Belgium one year. There are some assumptions, some caveats. I will for sure name them. But here, using this forecast of the average risk in Belgium and year forecast. Need to check myself. I think it was four hours before delivery here. It has generated a P&L of 161k per megawatt. You do also see on screen that some days are negative. So there are some losses. It's not always rosy. And maybe a big caveat is that on the 6th of April, we were on the right side in the model. But half the P&L was made on this day. So we're sensitive to our imbalance events. So it's just for you to be aware. I think you must have known for those or not. So on 6th of Easter Monday, we had some PV overshooting. And my reactivation for Germany were activated for minus 15k for 90 minutes. So creating really extreme imbalance prices in Belgium. So the assumption, the entry is an intraday market. That's the time of forecast creation. And the price which I took is an imbalance price. And one megawatt is traded on each non -neutral signal that has been created. Some caveats. So this P&L is, of course, a theoretical maximum. The execution will deliver less in reality for sure. Because you might need to do higher volume. The credit might be not there, etc. And then I have two more slides that's almost over to explain a bit of sensitivities between the P&Ls for what's playing in a role. So here I took the yearly P&L on a previous slide and I play a bit with the barometers we have. We can also do the same with the two. So we have different forecasts created at different times. So the time ahead to delivery is playing a role in a risk profile. And also in the middle, what I was trying to explain, your average risk, high risk or low risk. You can influence the guarantee and say, I just want to take position when I'm really sure the energy is really high. And in that sense, your accuracy will be higher, but the P&L will be lower. So this is a trade-off here for sure. But if you are ready to take high risk positions, there is some P&L on the way. So it's a way, let's say, for you to map the different P&L according to these two dimensions. And here, coming back to the regulation I was explaining to you. So ILIA is planning to allow passive balancing, but it's going to monitor it with a 60% rule. And so we did an analysis trial, are we compliant to this rule or not? So this is leading to this matrix. So you can see some colors and numbers. So it's from 0% to 100%. The goal was to be met at 60% on specific timestamps where the system balance was higher than 100 megawatts. And here, for most of the time horizons, up to one hour before delivery, we are meeting the criteria all year. So fine, but if you take on certain months, two hours, three hours, four hours, delivery, we are not meeting the criteria. It makes sense. The longer in advance, the more CQ is a forecast. And here, there's still a way to do that. And it will be actually my final slide. So what you can do, if you have like 10 megaflex, you can use the different forecasts we are proposing. So that you have to make and split four megs in a more secure way. Three meg moderate, three with aggressive and do like a blending, like a QV of the forecast we have. And yeah, to try to meet the forecast and maximize the P&R. And one thing is what you should then do when it's live. So this rule here, quite important. I think you should try it. Once you have a strategy, you should monitor it in real time and also adapt within months. So for example, if you are begging of a month, using a strategy on a top row, but you say, ah, no, I'm not meeting 60%. You should, within the month, reduce the risk and put more volume into the, let's say, more balanced strategies in order to meet the 60% and not have a warning from here. That's it for my side. Thank you very much. Bye. Thanks a lot, Vincent. We have time for a few more questions, maybe. So I wanted to see in the room if anyone has one for Vincent. Anyone? Yeah, in the back. Wait, coming up. Wim Vassa from Engie. In the previous presentation on the battery activity, you said if you have a size of, for example, 100 megawatt, you start to factor in a sort of kind of cannibalization on the impact. What do you think the passive imbalance strategy, if we would add participants all looking at NRV signals, pushing trades into the market? What do you think could be the impact of having those occasions? Yeah. And what's your thought on this potential evolution? That's a good question. Are you trying? Yeah, maybe this one. Yeah. So that's a good question. Something to definitely monitor on. I will tell you, I don't have the full answers, but I could maybe give some elements. I think here, what we are seeing on screen, so Elia is going, relaxing passive balancing, but first, before going two steps forward, one step backward, because today you have a lot of BRP taking passive balancing position. Now you have some monitoring, so it might at first reduce the passive balancing position. Then, depending on the setup, there are some rules, maybe new BRP are coming, so there will be like a saturation phase. But for sure, the passive balancing will have an impact on the imbalanced volumes that really has a requirement, so the activation price and maybe on longer term, like the capacity prices. But I don't have a firm view on how it will impact, but it will for sure have some impacts there. So, basically what we're afraid of is the same overshoot happening in Belgium that in Netherlands is currently happening. And Elia doesn't seem to be very worried about it, which I find strange, but we had discussions with the balancing department. The big question will be, is this going to affect how much AFR Elia will contract? Because if you see how much balancing capacity is now contracted, 117 or 171 megawatts, that's a discussion. I think 171 is the calculated need, and 117 is what you're contracting, which is very low. The assumption is that if you go for passive balancing, you can contract less capacity. But I don't think you can contract less capacity in Belgium. So, it might actually force Elia to contract more balancing capacity, which means that the merit order gets deeper. You get more moderate activation prices that we're already seeing, thanks to Picasso. So, yeah, the level of prices that we are seeing, thanks to Picasso, is also less attractive to chase with passive balancing. So, that will be a natural break on the impact of passive balancing. So, it's a combination of those factors. It's very difficult to say what it's going to turn out to, but there are mitigating circumstances. Nice. Any other questions? All right. Nice. Then I want to say one more round of applause for Vincent. Thanks a lot. Thanks. All right.