Google/Adsense doesn't track at the website level, they optimize at the user level. They are completely agnostic as to what website an ad appears on (provided that there is no spam content).

No testing required, the answer is three, because economics 101. If you limit the ads that can be shown you limit the participants in the auction. Reduced demand === lower price.

Finance is a competitive niche, that typically pays well. It is also a YMYL niche. The bar is very high. If you want to succeed you will need to get noticed by your advertisers. In finance that is typically straight forward, bring your advertiser customers.

Have you thought about affiliate marketing in addition to Adsense?

Google/Adsense doesn't track at the website level, they optimize at the user level. They are completely agnostic as to what website an ad appears on (provided that there is no spam content).

Every website is different even from the same category. They will be different in ad placements, UX, and the nature of users they attract. For example, an informational website from a niche may not see the same Adsense performance as that of a review website from the same niche. Even though the same user may visit each of those websites, his requirements from the websites will be different. Hence, I guess Google's machine learning algorithms will take website/webpage as one of the variables for their RPM prediction mechanism (based on which they serve ads).

No testing required, the answer is three, because economics 101. If you limit the ads that can be shown you limit the participants in the auction. Reduced demand === lower price.

Regarding auctions, I understand that the higher the demand, the higher will be the winning price (RPM). But I have a doubt. Since Adsense uses the second-price auction method, won't having a handful of 5 to 10 high bidders (say Group A) will have the same effect as having hundreds of low bidders along with the high bidders 'A' group. In a hypothetical scenario, in any particular auction, just 2 top bidders will alone be sufficient, because the second bidder will decide the price of the first. Please let me know if I am missing something here (I sincerely hope I am terribly missing something based on several threads on the same topic.)

Since Adsense uses the second-price auction method

Stop with the second price non-sense. It makes no difference to you.

Second price, means that bidder can enter a bid which is equal to the maximum value that he/she is willing to pay, with the understanding that bid price will only ever be realized if another bid is a penny less.

First price auction means that the price bid is the price realized.

If a bidder believes that they can get a lower price they will try to get the lower price, thus value of a bid in a first price auction will be less than the value bid in a second price auction. If the bidder looses, they will readjust next time around. In the end, over millions of auction the actual realized price between the two auction types will be the same.

There exists a market price and the type of auction will not impact it. In fact, both auction types are being used in the market now and establish that market price.

Hence, I guess Google's machine learning algorithms will take website/webpage as one of the variables for their RPM prediction mechanism (based on which they serve ads).

What are you talking about? There is no RPM prediction mechanism. Advertisers can select different categories that they can target, website is one of them. If the advertiser leaves the website blank their ad can appear on any website that meets their criteria.

I've found that it is best to enable everything. As NickMNS says, as soon as you start blocking advertisers or categories, you're reducing the number of bidders. Since it is an auction, fewer advertisers means less competition and lower prices. It rarely works to try to outthink a company with way, way more data than you'll ever have.

Assuming 100 advertisers compete for a slot, for Google to try each and collect data at the website level is going to take weeks or months.

No. Simply because Google also knows the performance of ads on other publishers' pages. And with personal ads enabled, Google matches the ad with the visitor, not with the site.

Simply because Google also knows the performance of ads on other publishers' pages

That is precisely my point. Google judges and delivers the ads with the (estimated) highest quality score (CTR X CPC) based on the data collected from thousands of websites. But what if a webmaster knows or can predict better than Google about the types (contextual/personalized/select categories) of ads that could work well on his site? He may be wrong, but is it not worth trying or testing?

If you know better than Google's algorithms, feel free to sell your ad space directly. If you want to stay within the Google eco-system, you can sign up to Ad Manager to better manage the ads. You may earn a higher CPM, but it comes at the cost of your time and effort.

Google AdSense is not the perfect solution, but it is simple and gives acceptable earnings for many publishers.