On-chain analysis will only give you information about what is happening "on chain" not "in market". In other words there may not even be a single bitcoin transaction or block being mined in a certain period while the market volume soars with traders buying bitcoin up on exchanges.

Even if you could find a link saying certain amount of coins were moved to exchanges or out of them you still can't make any conclusion about the market or speculate in any meaningful way.

You're a bit confused.
Before me others have already explained what on-chain analysis really mean: that's only good for reading transactions happening on the blockchain but all that is off-chain (exchange internal trading activity, L2 transactions etc.) is out of reach for the Blockchain Analytics company.
Therefore, no matter the volume, things don't change because of that.

On-chain analysis can also be used as an indicator which not guaranteed 100%, that's why you can see some on-chain analysis provider that they offer subscription, like some feature using their platform and the same time they have a free version.

So, I can say too that on-chain analysis can't be 100% reliable.

Other tools that I want to share from blockchaincenter.com

Dollar Cost Averaging
Satoshi Cost Averaging

the 3 things you can get that might(flimsy) aid assessing market support/value/price

1. the richlist.
    if you add up all the UTXO between 1btc-100btc and call that the collectors
                                           between 100btc-1000 call that the hotwallet
                                                 over 1000 call that the exchange cold wallet
                                               under 1btc call that the average joe users

and then periodically see the flow between the 4 groups. to estimate is the exchanges accumilating coins (more supply) or distributing out more coins (less supply)

2. coin-age
    take all the UTXO of the last 24 months. and work out the price vs the age.
    for instance. from january-now all UTXO created, moved when they were worth over $30k
                             february-may all UTXO created moved when they were worth over $40k
                               march -may all UTXO created moved when they were worth over $50k
    work out the percentage of coins at each increment of price..
    and you will have a good bases for the community underlying value support.
    basically the amount they are willing to sell at their lowest to break even.

3. look at the hashrate.
    calculate the hardware+electric of the hashrate based on the 5 main regions electric costs
      german/japanese asic farms
      german/japanese home hobbiest
      US asic farms
      US home hobbiest
      euro farms
      euro hobbiests
      chinese/icelandic farms
      chinese icelandic hobbiests


like (2) work out the incremental underlying break even value.
you will be able to spot if the price is higher then the cost. more mining in that group than buying
                                                      lower then the cost more buying in that group than mining
common sense is why mine slowly for sats in germany at over $40k.
when you can just buy it now in one lump for $32k
...
just to note.
with point (1) you cannot predict the day/time when a huge deposit into a exchange address converts into an active order on the market orderline.
 some people just use exchanges as a wallet and not a market

with point (2)(3) although common sense tells you people refuse to sell at a loss. meaning you can see some value support below the price. you cannot predict peoples level of greed as to the most they are willing to pay or sell for.    

meaning you cannot predict if they are a 'taker' seller (snipping off the top for instant sell cheap) or a 'maker' seller putting their coins at a stupidly high price that wont peak anytime soon